I want to talk about AI hacking. There are three aspects: people hacking AI systems, people using AI to hack other systems, and AIs hacking systems on their own, not even maliciously. The last two matter more than the first. The third is the weirdest and potentially the most destabilizing.

I started writing about this in 2021 and in my 2022 book, A Hacker’s Mind. Back then I said it was mostly still science fiction, but not stupid science fiction. We’re seeing it more and more, especially in the past year. The trends I’m describing don’t require a major breakthrough in AI. They’re starting to happen now, and they’ll happen more often as AIs get better at autonomous problem-solving. And it will involve far more than computer code.

AI will hack all sorts of social, economic and political systems. It will find vulnerabilities and exploit them. This is a difference in kind, not just degree. We’re not ready for it, and we need to think about the implications now.

This article is adapted from a keynote speech delivered at the DEF CON 34 conference in Las Vegas on August 7, 2026.

First, I want to broaden the term “hacking.”

The tax code isn’t computer code, but it is code. It’s a series of algorithms with inputs and outputs. It has vulnerabilities, which we call loopholes, and exploits, which we call tax avoidance strategies. Black hat hackers exploit them for profit. We call them accountants and attorneys.

What is a hack? One definition: “Something that a system permits, but is unanticipated and unwanted by the designers.” It’s subjective and implies novelty and cleverness. A hack follows the rules of a system but subverts its goals or intent, just as a software hack follows the logic of code while doing something its authors neither anticipated nor wanted.

All systems of rules can be hacked, including professional sports, customer reward programs, financial systems and politics. Hacks even work against our cognitive functions. A curved hockey stick is a hack, and we know the name of the man who invented it. Mileage runs were a hack. The filibuster was originally a hack in Roman times. Gerrymandering is a hack. Hedge funds and private equity are full of them.

A system is just a set of rules or norms, and the rules aren’t always formal. Even the best-designed rules will be incomplete, inconsistent or ambiguous. As long as people want to subvert a system’s goals, there will be hacks.

AIs are becoming hackers. After decades of slow progress, they’ve gotten very good at finding and exploiting vulnerabilities in computer code. They do this largely autonomously, and they’re getting better scarily fast. But the implications extend far beyond software, to the tax code, financial regulations and all sorts of social, economic and political systems.

There are two distinct issues. First, someone might instruct an AI to hack one of these systems, feeding it the world’s tax codes or financial regulations to find profitable loopholes. Second, an AI might hack a system on its own, inadvertently. Both are dangerous, but the second is worse because we might never know it happened.

Now, reward hacking. AIs don’t solve problems the way people do. They stumble on solutions we never anticipated, and some subvert the intent of the system. They don’t share our context and don’t necessarily understand the implications, norms and values we take for granted.

Reward hacking is when an AI achieves a goal in a way its designers neither wanted nor intended.

Early pre-LLM examples are telling. In a two-person soccer simulation, an AI learned that kicking the ball out of bounds forced the opposing goalie to throw it in and leave the goal undefended. In a stacking simulation, an AI learned to flip a block upside down rather than stack it. In an evolution simulation where the goal was to cross a distant finish line, an AI made itself tall enough to fall over the line instead of growing longer legs or stronger muscles.

These are all hacks, the result of poorly specified goals or rewards. Today, they’d come from bad prompting.

The problem is universal. One researcher tried to teach a robot vacuum not to bump into things. Instead, it learned to drive backward, because there were no sensors in the back.

Any capable AI will find hacks wherever the rules have problems, inconsistencies or loopholes that still produce an acceptable solution. We see a hack because we understand context. The AI doesn’t, at least not in the same way.

This problem shows up often in myth.

In the story of King Midas, the god Dionysus grants him a wish: everything he touches turns to gold. Midas ends up starving when his food, drink and daughter all turn to gold. It’s a specification problem. He programmed the wrong goal.

In the Golem of Prague, a ghetto animates a clay statue to guard its residents. It guards them past all reason until they figure out how to turn it off. This is a guardrail problem. The villagers never limited the golem’s actions.

More generally, think of this as the genie problem. Genies are precise about the wording of wishes and can be maliciously pedantic when granting them. There’s no way to outsmart a genie. Whatever you wish for, he can grant it in a way you wish he hadn’t.

The reason is that in human language and thought, goals are always underspecified. We never include every caveat, exception and proviso, and we never close off every avenue for hacking. Any goal we specify will be incomplete. That’s largely fine in human interactions, because people understand context and usually act in good faith.

If I asked you to get me coffee, you’d pour a cup from the nearest coffee pot or walk to the corner shop. You wouldn’t bring me a pound of raw beans, buy a coffee plantation or snatch a cup from a stranger’s hands. I wouldn’t have to specify any of that. You’d just know.

Similarly, if I asked you to develop a technology that turns things to gold on touch, you wouldn’t build it to starve the person using it. You’d just know.

In 2015, Volkswagen was caught cheating on emissions tests. Human engineers programmed a regular computer to detect testing and behave differently during it. The cheat went undetected for years because it’s hard to tell what software is doing.

If I asked you to design engine control software that maximizes performance while passing emissions tests, you wouldn’t design it to cheat without realizing you were cheating. An AI would, because it will think “outside the box” when it has no concept of the box.

Unless programmers specify that the system must not behave differently during testing, an AI might come up with the same hack. The programmers would be satisfied, the accountants would be ecstatic, and no one would know until someone audited the code.

We can now constrain the goal to avoid that particular hack, but programmers won’t anticipate others. The lesson of the genie is that there will always be hacks programmers don’t anticipate.

We’re building AI genies: systems that take broad instructions and figure out how to carry them out, sometimes in ways we didn’t expect or intend. That’s the whole point of vibe coding. You underspecify what you want and let the AI work it out. Sometimes that produces genie behavior.

We can’t completely specify goals to an AI, and AIs can’t fully understand context. More precisely, current AIs have a shallow grasp of concepts. If we asked a current LLM to get me coffee, it would almost certainly do the right thing.

I worry about goals that aren’t well represented in training data, and about hacks that aren’t obvious, whose effects are subtle, like the Volkswagen hack.

Recommendation engines are an early example. They push people toward extreme content. No one programmed that. It emerged because that’s what people respond to. It didn’t take a bad actor. A basic automated system with feedback found the hack on its own.

We’re now seeing a second generation with today’s AI coding systems. AI researcher Simon Willison asked Fable to track down a stray scrollbar in a web app. When he returned, it had opened browsers, written its own screenshot tools, edited the framework’s internal templates to make the bug appear on cue and set up a local web server to collect measurements. It found the bug, but it also did many things he never asked for. Another researcher gave Fable a complex task, and it tried to hack its own memory limits because it needed more storage.

A new OpenAI model did something similar when it hacked into Hugging Face. It was being benchmarked on its ability to turn vulnerabilities into exploits. It decided the best way to reach its goal was to break out of its containment, access the internet and break into the Hugging Face network, because it thought the answers would be there.

Today’s LLMs are acting like genies, especially when powerful models are combined with sophisticated harnesses.

There are two kinds of genies. The Dionysus genie does the wrong thing. It understands your prompt literally but not correctly, and returns a mess you never intended. The Golem genie does the right thing but tramples everything in its path to get there. The paperclip maximizer is a Golem genie.

Consider some examples. Ask an AI to deal with spam calls, and a Dionysus genie might change your phone number. Ask it to book a particular flight, and a Golem genie might hack the airline database because the flight is full. Ask it to get a refund for an online purchase, and a Dionysus genie might draft a legal threat on fake letterhead and send it to the company. Ask it to buy tickets to a popular concert, and a Golem genie might discover the ticketing system is rate-limited and spin up thousands of cloud servers to hammer it. You’d know if some of these things happened, but you might never realize if others did. More detailed prompting won’t fix this. The lesson of the genie is that no amount of detail is good enough.

AI researchers aren’t surprised by any of this. Many are working on defenses against reward hacking and subtler forms of genie behavior. Some of this falls under “value alignment”: How do we teach AIs context? And how an AI satisfies a goal matters as much as whether it does. I want to build a benchmark to measure genie-like behavior, so we know where models stand and whether alignment is improving.

Think beyond LLMs and computer code. Consider specialized AIs built to find loopholes, and hacks of more general systems: the tax code, financial regulations and democratic procedures.

Financial systems are the most likely first target, because their rules are designed to be algorithmically tractable. High-frequency trading is itself a hack. Imagine giving an AI all of the world’s financial laws, plus all the news and financial data, and the goal of finding profitable loopholes. I suspect large investment banks are already doing this. The results will be novel tricks that are legal but unintended and unwanted, some of which might break the financial system.

Tax loopholes are the next likely target, for the same reasons. For a while, AIs will identify potential loopholes, and human hackers, namely attorneys and accountants, will decide which are feasible.

Further out, expect hacks of regulatory, legislative and procedural systems, and of the broader economy.

Hacking is as old as humanity. We’re creative problem solvers and loophole exploiters. We manipulate systems to serve our interests and strive for more influence, power and wealth.

Still, no human pursues their interests without constraint. Even sociopaths are limited by the complexity of society and their own contradictory impulses. They worry about their reputation and punishment, and they have limited time. These human qualities have limited hacking.

Hacking changed as everything became computerized. Computers are uniquely hackable because of their formalism and complexity. Today, nearly everything is a computer, from cars and appliances to phones. Our social, economic and political systems, including taxation, finance and elections, all involve computers.

Until recently, hacking was exclusively a human activity. Finding a novel hack required expertise, creativity, time and luck.

AI hacking will change everything again. AIs aren’t bound by the constraints that limit people. They’ll think like aliens, and they’ll change hacking’s speed, scale, scope and sophistication, sometimes at our direction and sometimes on their own.

Speed is the easiest to see. A creative process that takes a human months or years could be compressed into hours or seconds.

Scale is next. Once AI systems start discovering hacks, they can exploit them at a scale we’re not ready for. We already see this with AI-generated slop, which is rampant on social media. AI bots post around the clock, sometimes overwhelming real human discussion. It’s hard to know how much of social media is still people talking to people. What looks like boisterous political debate may be bots arguing with other bots, artificially shaping what we think and what we think others think.

Scope makes hacks more dangerous. AIs already make decisions that affect our lives, decisions we once thought belonged only to humans. As AI systems grow more capable, society will hand them more decisions, and more important ones. Hacks of these systems will do more damage.

Finally, AI hackers are more sophisticated. They can hold more variables in working memory and generate more complex hacks. What might happen if you fed an AI the entire U.S. tax code? Would it figure out, without being told, that it’s smart to incorporate in Delaware and register a ship in Panama? Consider the Double Dutch Irish Sandwich, the strategy Apple and Google used to avoid paying U.S. taxes. It involves U.S., Dutch and Irish law and an offshore tax haven. How many loopholes will an AI find that we don’t already know about? Dozens? Hundreds? Thousands? We have no idea.

These hacks will largely be perpetrated by the powerful against the rest of us. AI is fundamentally a power-enhancing technology, and so is hacking. The more power you already have, the more a hack gives you. If you or I found a tax hack, we might make a few thousand dollars. If a large investment bank found one, it and its clients could make billions.

Our societal systems for dealing with hacks were built when hackers were human, and they reflect the pace of human hacking. We have no system of governance that can handle hundreds, let alone thousands, of newly discovered tax loopholes. We won’t be able to recover from an AI finding unanticipated but legal hacks of financial systems, operating at AI speed, scale, scope and sophistication. Hacking becomes a problem society can no longer manage.

Now, defense. AIs that find software vulnerabilities help defenders as well as attackers. Vulnerabilities can be patched, removing them permanently, or turned into signatures that security systems use to detect and block attacks. In theory, AI could identify and patch every automatically discoverable vulnerability before a product ships. This is rapidly becoming routine in software development. We call it vulnops. We can imagine a future where accidental software vulnerabilities are a thing of the past.

Deliberate vulnerabilities are different. There will be an arms race. AI will craft increasingly obscure back doors, and AI will find them. It becomes a contest between the speed of creation and the speed of patching, and between obfuscation and reverse engineering.

The same logic applies to broader systems. Yes, AI hackers might find hundreds of vulnerabilities in the existing tax code. But the same technology can evaluate potential loopholes in any proposed tax law or ruling. A legislator, a watchdog group or a reporter could take the text of a bill and identify every exploitable loophole before it passes. The difference is that, unlike software, knowing about a tax loophole doesn’t mean it gets patched. In theory, though, we can fix them before the rich and powerful exploit them. So maybe defense will come out ahead here, too.

That requires resilient governing structures that can respond quickly. It won’t help if updating the tax code takes years, or if a legislative hack becomes so entrenched that it can’t be fixed politically. Maybe a lobbyist put it there on purpose. Maybe it was discovered after the law passed, and now lobbyists want to keep it. Microsoft can patch the vulnerabilities it finds. Patching is much harder when legislatures have to do it. We need society’s rules and laws to be as patchable as our computers.

This is a hard problem of modern governance, and it goes well beyond AI. It’s the problem of building governing structures for the speed and complexity of the information age.

The umbrella idea is trust. What do we need for AI systems to be truly trustworthy? Capability isn’t enough.

AIs won’t be trustworthy if the companies controlling them spy on us or subtly manipulate us. I call this the friend vs. tool problem. Will we have AI agents, or AI double agents?

AIs won’t be trustworthy if they’re insecure. How do we know they haven’t been manipulated? Almost all AIs operate in adversarial environments, so someone will benefit from hacking them. Unless we know they’re secure, we shouldn’t trust their output.

And AIs won’t be trustworthy if they act like genies. We need to know they’ll perform tasks the right way, not just perform the tasks we ask for.

I’ve written extensively about the need for “integrous” AI, including how to think about the problem, what a research agenda looks like and what good laws and policies look like. I predict integrity will be the key security problem of this decade, just as confidentiality was the key problem of the last one. This applies to both AI and the internet of things.

The big shift is agency, which I think of as Web 3.0. Computers now affect the physical world directly and with growing autonomy. Sensors go from data sources to attack surfaces, and actuators go from outputs to attack capabilities. An integrity failure no longer means bad spreadsheet data. It means the wrong dosage, the wrong valve position or the wrong braking distance.

Integrity isn’t one thing. I break it into input integrity, processing integrity, storage integrity and contextual integrity. Each is complex and important, and each deserves its own talk. I’ve written about them, and I hope you’ll read it.

Much of this isn’t really an AI problem. It’s not a technical problem so much as a capitalism problem or a democracy problem.

The failures of late-stage capitalism and democracy in the information age both stem from the rich and powerful getting too good at hacking our social, economic and political systems. This is the core argument of my 2021 book.

Seen this way, AI is exacerbating existing problems. Human inefficiency has largely kept those problems in check, and that won’t work much longer.

So consider this a call to action. Technically, we need to understand how AIs hack systems, build systems that are resilient to hacking and build AIs that won’t act as genies. Politically, we need to learn to govern and regulate at the speed of technology. Otherwise, market failures will overwhelm our ability to govern them.

I said hacking is as old as humanity and that AIs will overwhelm our ability to deal with it. The solutions aren’t specific to AI, but AI makes them more urgent. We need to understand them, and we need to implement them.

The post AI doesn’t have to go rogue to break the world appeared first on CyberScoop.

Three men concocted an elaborate, yearslong scheme that defrauded companies of at least $129 million by using AI to impersonate executives and lawyers, tricking targeted victims into transferring large payments to criminal-controlled accounts, the Justice Department said Thursday.

Jacob Soall, Sharon Hazan and Guy Sheetrit allegedly used AI to impersonate the voices and images of chief executive officers at companies and lawyers at major international law firms. The trio are accused of targeting thousands of victims between July 2024 and February 2026, according to court records.

Victims include a U.K.-based financial tech company that lost almost $11.6 million, a Florida-based pet food company that lost $5.2 million, a Missouri-based hospital system that lost $4.8 million, a Massachusetts-based mutual fund company that lost $4.6 million, a California-based hospital system that lost nearly $3 million, a New York-based investment company duped out of $2.8 million, and a Rhode Island-based bank that transferred nearly $2.7 million.

The trio’s scheme typically started by targeting accounting staff, pretending to be the CEO, and requesting help on a financial transaction, officials said. Using AI to create deepfakes pretending to be CEOs, the crew allegedly directed employees to contact attorneys who they would also impersonate to trick the employee into thinking the financial transaction was legitimate.

Prosecutors said the men and their co-conspirators identified and targeted employees who had sufficient authority to authorize large wire transfers. 

The scheme typically began with an email from an e-signature company, wherein the impersonated CEO claimed they were working on a “confidential financial transaction” that required the victim employee to sign a non-disclosure agreement before contacting the purported attorney at a victim law firm, according to an unsealed indictment.

Soall, Hazan, Sheetrit and their co-conspirators allegedly impersonated attorneys at some of the 50 largest law firms globally. 

“Super intelligence technology offers tremendous potential, but the public should recognize that there are some who seek to use this emerging technology to pursue criminal goals,” Jamie McDonald, U.S. attorney for the Southern District of New York, said in a statement. 

Prosecutors said the wire transfers were sent to bank accounts controlled by co-conspirators around the world, including the United States, Spain, Portugal, Greece, the United Arab Emirates, Singapore, Austria, Slovakia and China. The crew allegedly converted the funds to cryptocurrency and distributed the illicit proceeds amongst themselves and their co-conspirators.

Soall, a 41-year-old from Marbella, Spain, was arrested in Spain on June 17. He allegedly primarily impersonated attorneys from the United Kingdom, reading from a handwritten script designed to dupe targeted victims.

Hazan, a 54-year-old from Israel, was arrested in Spain on Aug. 5. He allegedly boasted that he had been laundering money since 1994, including more than $400 million in the past three years, and was previously convicted in the Southern District of New York of conspiracy to commit bank fraud in 2008.

Sheetrit, a 42-year-old from Israel and the United Arab Emirates, was arrested in the United Kingdom on Sept. 7 and allegedly laundered funds from the scheme.

The trio are charged with conspiracy to commit wire fraud, wire fraud, two counts of aggravated identity theft, money laundering conspiracy and two counts of money laundering. The seven counts carry a maximum sentence of 104 years in prison.

“The alleged scheme orchestrated by Soall, Hazan, and Sheetrit demonstrates how the use of advanced technology can be exploited to commit large scale fraud,” James C. Barnacle Jr., assistant director in charge of the FBI New York field office, said in a statement.

“As criminals continue to weaponize emerging technologies, the FBI remains committed to rapidly adapting our capabilities to identify, investigate, and disrupt those who seek to harm New Yorkers and businesses operating in our communities.”

After initial publication of this story, Sheetrit’s attorneys, Sharon Nahari and Shay Shorer, emailed a statement to CyberScoop saying their client “has no connection to this matter and has had no involvement in it whatsoever.”

“Mr. Sheetrit is a businessman with extensive expertise and professional experience in cybersecurity and technology,” wrote Nahari and Shorer. “In the course of his work, he provides professional advisory services to a range of clients. Working in cybersecurity, possessing specialist expertise or providing advisory services cannot, in themselves, establish complicity in or involvement in the alleged conduct.”

“Our client maintains his complete innocence and, like everyone else, is entitled to the presumption of innocence,” the statement continued. “We are confident that, once the facts have been established and the evidence fully examined, it will become clear that there is no basis for holding him criminally responsible.”

UPDATE: Nov. 6, 2026; 8:25 p.m.: This story has been updated to include a statement from attorneys representing Guy Sheetrit.

The post Authorities accuse trio of using AI deepfakes to defraud $129M from businesses appeared first on CyberScoop.

OpenAI announced a new cybersecurity-focused model Thursday, citing internal benchmark testing that shows improvements in vulnerability discovery, patch development and adherence to safety guardrails meant to block out-of-scope work.

GPT-6 Cyber will first be available to a limited number of customers in OpenAI’s Daybreak Red program, which uses more advanced or nonpublic cybersecurity models for offensive red teaming.

As a preview of the model’s capabilities, OpenAI officials said GPT-6 has found at least 81 “novel” vulnerabilities in the Firefox web browser, developed exploits for 948 Linux kernel vulnerabilities and was able to develop and validate a critical remote code execution vulnerability in Brotli, a popular open-source data compression tool. All the vulnerabilities were reported to owners and patched.

OpenAI officials told reporters Wednesday that the new safeguards include additional identity and account security requirements, more frequent check-ins from the model to OpenAI when a request is made to perform high-risk behaviors, and additional monitoring for “misalignment” or rule-breaking.

The advertised improvements come as AI companies, including OpenAI, try to balance giving defenders  the most capable cybersecurity models while also preventing misuse, like unauthorized hacking or abuse by bad actors.

Eric Wallace, a researcher at OpenAI who focuses on training, evaluation and sandboxing for new models, said part of the training for GPT-6 Cyber included a simulated evaluation meant to mimic the Hugging Face incident, where the company’s models escaped internal sandbox testing environments and hacked into part of a major online AI code repository.  

“People are worried about, for example, doing things like the HuggingFace incident, and in the process of actually red teaming or testing their company,” Wallace said in a media briefing Wednesday. “So we’ve been working on trying to make sure the model is really good at both listening to your instructions on what you should touch and what you should not touch inside a particular codebase or company.”

Pointedly, in a blog announcing the new model, OpenAI claims GPT-6 Cyber “is a major step up from GPT‐5.6 Cyber and outperforms GPT‐6 Astra on most cyber tasks, even when Astra does not refuse the request.”

Wallace said since the summer, when OpenAI’s 5.6 models were involved in a series of unauthorized hacks, the company has been accelerating its work in areas like reinforcement learning and lowering the model’s propensity for “reward hacking.”

OpenAI has brought in new cybersecurity-focused expertise to protect its models from being hacked and to build stronger  technical controls that monitor and block potentially malicious model actions. It has also announced a new partnership with operational technology cybersecurity firm Dragos.

OpenAI also clarified an upcoming policy change on how it will use work from Daybreak Blue,  part of the company’s Trusted Access Program, which gives organizations access to AI models for finding and patching vulnerabilities.

Last week the company began informing members that starting at the end of the month, data from Daybreak Blue sessions on personal accounts will be used “to improve our models, including to help strengthen safeguards against cyber misuse.” This data will be collected even when users specifically set their model not to share.

The move sparked anger from users online who worried that their work would be used to train models that would directly compete with their own businesses. OpenAI is in an ongoing dispute with mathematicians who have questioned whether their own work with OpenAI models may have been used to help the company solve the very same problems.

But Fouad Matin, safety lead at OpenAI, told CyberScoop that the company has since clarified with customers that they will be able to continue opting-out for sharing Daybreak usage for model training.

Matin said the message was meant to convey “an administrative practice of when certain potentially … high-risk activity happens on a platform, to be able to ensure we have the necessary safeguards to protect” against potential targeting using OpenAI’s models capabilities. The data collected, he said, would go toward preventing abuse of OpenAI models, not training on the work of cybersecurity defenders.

“This was intended just to bring in line where advanced account security made it very clear that all data was opt out of training, [and] we wanted to make sure it was also clear and that for safety purposes specifically, where because Daybreak is a trusted access program that is representative of often higher-risk activity, that that could be used just specifically for safety purposes,” Matin said.

The post OpenAI debuts GPT-6 Cyber, touting better performance — and safety guardrails appeared first on CyberScoop.

Democrats’ big win on Election Day, which saw them take control of both the House and Senate, has ramifications for the Cybersecurity and Infrastructure Security Agency, artificial intelligence, surveillance policy and much more.

Here’s what lawmakers and cybersecurity experts have to say about the short-term and long-term prospects on Capitol Hill.

Leadership plans ahead

Maggie Hassan, D-N.H. is in line to take the gavel on the Senate Homeland Security and Governmental Affairs Committee, with Sen. Gary Peters, Mich., retiring.

“Our government’s number one job is to keep people safe, and the Senate Homeland Security and Governmental Affairs Committee must work to ensure that our country remains safe, secure and free from all kinds of threats,” Hassan said in a statement to CyberScoop. “That work involves protecting against terrorism and fentanyl trafficking, but it also includes guarding against emerging threats like artificial intelligence security risks and strengthening our cybersecurity.”

“The Trump-Vance administration has undermined our security on these fronts, including by diverting resources away from combatting fentanyl trafficking and gutting CISA,” she continued. “We need to reverse these cuts, confront these challenges and put public safety first.”

Rep. Bennie Thompson, D-Miss., is poised to return once again as chair of the House Homeland Security Committee, said Democrats are taking over at a “pivotal time for CISA,” with it looking to add staff after cutting back by nearly one-third.

“First and foremost, we will aim to work with CISA to expedite its critical re-staffing effort and to ensure it has the expertise necessary to support stakeholders as they defend against the scale and velocity of AI-enabled cyber-attacks,” he said in a statement to CyberScoop. “Advances in AI can help defenders, but it will also accelerate the consequences of deferred investments in security and CISA must be well-positioned to help critical infrastructure owners and operators rapidly improve their cybersecurity posture.”

“We will prioritize getting resources straight into the hands for defenders — by extending the State and Local Cybersecurity Grant Program and improving support of under-resourced sectors, especially water,” he continued. “We will work to ensure CISA re-establishes its cross-sector coordination capacity, which was diminished when the Trump administration- terminated CIPAC, by making sure the Anchor-CI is eventually up and running and meets the needs of stakeholders. Finally, we will make sure CISA is well-positioned to support the secure development and deployment of AI.”

What happens in the lame-duck session

Before the GOP relinquishes control of the House and Senate to Democrats, though, there’s a lame-duck session ahead. 

Long-term renewal of the Cybersecurity Information Sharing Act of 2015 (CISA 2015) is a bipartisan and industry priority, with it set to expire Dec. 11. Current Senate Homeland Security and Governmental Affairs Chairman Rand Paul, R-Ky., has been the biggest obstacle to reauthorization, but he could get “rolled” by GOP leadership on an extension, said Michael Garcia, a former government cyber official and vice president of the cybersecurity practice at Monument Advocacy. Possible vehicles for CISA 2015 to hitch a ride include an omnibus funding bill or the annual defense policy bill.

The Department of Homeland Security, and its subagency CISA, are currently operating on a continuing resolution that’s also set to expire Dec. 11, where lawmakers could temporarily extend CISA 2015 if a long-term solution can’t be reached in the lame duck. Another continuing resolution before Dec. 11 in lieu of an omnibus spending bill is an option for the lame duck. Both a CR and omnibus spending bill would theoretically include DHS funding, although the agency’s funding has been made more difficult in the past by the presence of Immigration and Customs Enforcement within the department.

Republicans have talked about wanting to get the fiscal 2027 National Defense Authorization Act done in the lame duck. Besides the possibilities of CISA 2015 being included in the bill, Nicholas Leiserson, senior vice president for policy at the Institute for Security and Technology, said the Senate version of the legislation has a pair of very noteworthy cyber provisions: a “very broad” authorization for contractor-owned, contractor-operated offensive cyber operations, and a requirement that every artificial intelligence product sold to the Defense Department have an AI bill of materials.   

Additional leadership shuffle

While Paul is losing his gavel, he’s shown he can be a disruptive force on passing cybersecurity legislation through the Senate in the minority. That makes what he does next possibly determinative for cyber bills, Leiserson and Garcia said. “If he leaves HSGAC, let a thousand cyber bills flourish!” Garia quipped on social media. Paul could opt to become the top Republican on the Senate Health, Education, Labor and Pensions Committee instead of HSGAC.

Nebraska Sen. James Lankford would take over for Paul as top HSGAC Republican in that scenario, and an industry source said he’s viewed as a supporter of CISA 2015.

On Senate Intelligence, the panel’s leadership goes to a lawmaker who has placed a big emphasis on cybersecurity in his congressional career: Mark Warner, D-Va. 

On the House side, while Rep. Andrew Garbarino, R-N.Y., moves to the minority on the Homeland Security panel, he has proven willing to collaborate with Democrats on cybersecurity issues. Rep. Delia Ramirez, D-Ill., is poised to take the gavel of the panel’s cybersecurity subcommittee, and Rep. Vince Fong, R-of California looks set to become the subcommittee’s top Republican and has his own history of bipartisanship. Fong would replace current subcommittee chair Andy Ogles, the far-right Tennessee Republican who lost in the primaries.

One thing to watch in the House, Leiserson said, is how the House looks to tackle AI — through existing committees, a select committee or some other structure. Democrats haven’t put forward a unified front on AI guardrails yet.

With AI models uncovering as many vulnerabilities as they are, “highly capable frontier models are going to drive a lot of discussion” about AI and cybersecurity, said Craig Albright, the Business Software Alliance’s senior vice president for U.S. government relations.

Appropriations, oversight, legislation

If the NDAA or a CISA 2015 extension don’t pass in the lame duck, they’d be priorities in 2027, too.

“CISA 2015 has to happen, and if we can’t get it done, then we can’t move to the next thing,”  said Henry Young, BSA’s policy director. “Once we get it done, there’s an open question about which of these priorities that we’ve talked about gets to be number one. Is it post-quantum cryptography? Is it government IT modernization? Is it frontier AI cybersecurity systems? We don’t know that yet, and we can’t know until we pass this law.”

How damaged the agency CISA is by Trump administration’s budget cuts is something Democrats are expected to scrutinize closely. Another top budget issue is what to do with the expired State and Local Cybersecurity Grants Program housed at DHS, since even if it’s reauthorized, it will require appropriators taking steps to invest in it.

Democrats could take a look at a variety of other Trump administration changes at DHS and CISA, on topics like election security, the Cyber Safety Review Board, the Joint Cyber Defense Collaborative and more. 

More broadly, Leiserson said an area of Democratic cyber oversight is likely to be the Department of Government Efficiency that cut cyber personnel at agencies at and beyond CISA, and “particularly the integrity of systems that DOGE touched when they were given very high privileged access to federal systems.”

If the White House puts forward a permanent nominee for director of CISA, Democrats would be poised to vet that person. They also would have a say in who leads the Office of the National Cyber Director, should the incumbent, Sean Cairncross, depart.

So far, Democrats haven’t put forward a unified posture on expired surveillance authorities known as Section 702, although both Warner and incoming House Intelligence Chairman Jim Himes, D-Conn., have fought in the past for its renewal. The surveillance program is operating similarly still due to a court order, but that order only lasts through next spring.

The post What Democratic control of Congress means for cybersecurity appeared first on CyberScoop.

CrowdStrike announced Monday it has signed agreements to acquire two artificial intelligence security companies, Fiddler AI and Dreadnode, in deals announced at its Fal.Con Europe conference in Barcelona.

CrowdStrike did not disclose what it paid for the companies. A report in the Wall Street Journal listed the price for Fiddler at $400 million, and Dreadnode at $95 million. 

Fiddler, which specializes in AI agent observability and explainability, will become Falcon Observe on the company’s Falcon platform. Dreadnode builds automated security testing for AI systems, checking models and agents for weaknesses before deployment.

The acquisitions address a gap that has widened as companies adopt AI agents: the software can hold valid credentials and approved tools yet still make bad decisions, ignore instructions or claim work it never completed. Traditional security testing, built for code that behaves predictably, does not reach the model layer or the agent’s reasoning process.

Fiddler’s technology works after deployment. It evaluates agents before they reach production, traces decisions and reasoning across agent systems, flags drifting behavior and made-up answers, and creates records that can be used in audits. It works with common agent frameworks including LangGraph, Amazon Bedrock and Google ADK.

“When a regulator or your board asks what an agent did, saying ‘the model did it’ is not an answer,” said George Kurtz, CrowdStrike’s chief executive and founder, describing Fiddler as “the flight recorder for every agent.”

Dreadnode works earlier in the process. Its platform runs more than 100 known attack methods, including jailbreaks and prompt injection, against models and agent systems. It ranks problems by severity and maps them to widely used security standards, such as the OWASP Top 10 for large language models and the NIST AI Risk Management Framework. Dreadnode Chief Executive Brad Palm said his company was built on the idea that “the people best equipped to find those failures think like attackers.”

The announcements come alongside new Falcon Guardian capabilities that protect agents while they run. In a separate announcement, Kurtz introduced what the company calls the AI Kill Switch, a control that lets humans stop any AI agent. Writing on X, Kurtz said the switch “lives outside the model, at runtime, where an agent can’t argue with it, rewrite it, or wait it out.” He argued that the ability to stop an agent is what allows companies to run them faster: “The more certain a company is that it can stop any agent, the more agents it can put to work.”

Kurtz framed the shift as a move from security to safety. “For decades, security meant keeping the adversary out,” he wrote. “Safety is the bigger opportunity: letting AI run at full speed, inside the business, with a human in control.”

Both products are expected to be generally available in CrowdStrike’s fourth quarter of fiscal 2027. Both deals are expected to be paid mostly in cash, with part of the payment made in company stock that will be given out over time as conditions are met. They are expected to close in that quarter, pending standard approval steps.

The post CrowdStrike moves to acquire Fiddler and Dreadnode, adding AI observability and testing to its security platform appeared first on CyberScoop.

A China state-backed threat group is using ARTEX, the automated agentic AI penetration testing framework, to achieve remarkable speed, scale and success in a campaign that already compromised about 140 sensitive targets in 11 days, according to CrowdStrike. 

The threat group, which CrowdStrike tracks as Vapor Panda, targeted more than 200 organizations in at least 14 countries since Oct. 17, the security company said in a threat report Thursday. Targets and confirmed victims include foreign ministries, a European country’s presidency and prime minister’s office, defense and intelligence agencies; 19 U.S.-based organizations in the pharmaceutical, semiconductor, defense, nuclear and financial sectors; and telecom providers in the United States, Belgium and the Philippines. 

CrowdStrike told CyberScoop about 70% of Vapor Panda’s targets were confirmed compromised or had evidence indicating moderate confidence of a successful intrusion. 

Vapor Panda’s use of ARTEX, paired with DeepSeek, allowed it to accomplish more tasks with shocking speed — requests occurring as often as every 10 seconds — and pursue multiple attack paths concurrently. 

“This is the first time that we’re seeing an agentic adversary nation-state threat actor at this scale, and I think that this is probably, unfortunately, something we’ll see more of,” Adam Meyers, senior vice president of counter adversary operations at CrowdStrike, told CyberScoop.

Hands-on-keyboard activity tracked by CrowdStrike for the past 15 years typically involved up to 40 commands per hour. “Now, with ARTEX, we could see it exceed 1,500 commands an hour,” Meyers said.

“If you diversify that across 200 targets, it costs them very little to run a mass-exploitation campaign and drop a websell,” he added. “They can hit more targets in a faster period of time.”

Vapor Panda’s broad-access campaign, which remains active, has targeted think tanks and government agencies spanning defense, justice, finance and diplomacy.

“I would characterize it as a diverse operational mission that tends to focus on sectors and geographies that Beijing considers economically or politically important,” Meyers said.

The active campaign attributed to Vapor Panda is separate from the ARTEX activity Microsoft Threat Intelligence attributed to Storm-0940 earlier this month, including a supply-chain attack targeting ConnectWise customers. 

Adversarial use of AI is relatively new, yet also quickly becoming the new normal, Meyers said. “We’re scrambling because we’re seeing so many of these agentic adversaries.”

Agentic AI use for malicious purposes is growing month-over-month at a rate of almost threefold. CrowdStrike observed roughly 45 agentic adversary attacks in September and it’s already at 126 for October, and climbing.

Adversarial use of agentic AI isn’t an intractable problem, but rather another reminder that organizations need visibility into their endpoints, identities and cross-domain capabilities, Meyers said. 

These attacks aren’t showcasing anything new, “it’s just doing it at a speed and scale that we’re unaccustomed to,” he said. “It’s not hopeless, but you’ve got to take action. You can’t put your head in the sand.”

The post China-backed threat group used agentic AI pentesting tool to compromise 140 sensitive targets in 11 days appeared first on CyberScoop.

Even OpenAI is admitting there is a lot of AI slop in the security space. 

An executive for the frontier AI company said Tuesday that AI-generated vulnerability reports are overwhelming open-source maintainers, leading some to stop accepting submissions as the workload exceeds their capacity to review them.

Michael Aiello, OpenAI’s cyber executive in residence, described the problem during a question-and-answer session Tuesday at the DataTribe Challenge in Washington, D.C., when he was asked how he viewed OpenAI’s Patch the Planet program versus open-source maintainers banning AI-based submissions because of concerns about quality and clarity. 

“It’s an ecosystem problem where that [maintainer] is just not going to respond to hundreds and hundreds of AI-submitted vulnerabilities,” Aiello said. 

He contrasted the volume against his company’s own program, which partners with security research organizations to identify vulnerabilities, develop fixes and help bring them into the affected projects. According to the company’s website, 661 patches have been developed and tested, with 458 of them accepted by maintainers for inclusion in the projects they oversee. 

However, Aiello said those involved with Daybreak often coordinate with maintainers, as opposed to just sending a message into GitHub and hoping for the best. 

He further described the broader burden on maintainers, who are often tasked with running through hundreds of AI-generated reports to determine validity, the security implications and evaluate the proposed fix.

“In most cases, they’re likely real bugs of some degree,” he said. “It’s, ‘Are they critical bugs?’ And are they bugs worth the time of the maintainers?…The result is they’ve just stopped taking” submissions.

Aiello further laid out advice for those that do create vulnerability disclosures for open-source maintainers, including more work in establishing the quality of the finding, writing an understandable report and allocating resources to coordinate disclosure once a report is filed.

“Make sure they’re really high quality, make sure it’s consumable and readable by a single person, and then spend your own resources coordinating vulnerability disclosure,” he said. “You need to now coordinate the vulnerability disclosure across the whole ecosystem, which is itself another project.”

Aiello used OpenAI’s work on the recent PHP vulnerabilities as an example of that coordination. He said a partner research organization discovered the flaw, and OpenAI worked with the PHP Foundation to validate it, develop a fix and coordinate its release.

“You have to work with response partners, and you have to work with the operators to actually issue the patch,” he said.

The process involved coordinating with a broad group of organizations, extending the work from technical review through patch release. Aiello acknowledged difficulties in carrying out the entire process.

“Now, did it go perfectly? No,” he said. “We’re learning as we do this over time.”

The post Before you file that AI bug report, OpenAI has some advice appeared first on CyberScoop.

Anthropic said its AI systems could drastically reshape the software and cybersecurity sectors in the “near term,” while also laying out a litany of cybersecurity and regulatory risks facing the company as it prepares to go public.

In a filing with the Securities and Exchange Commission ahead of its initial public offering, the company said its technology could transform entire industries and the economy. Already, cybersecurity-focused models by Anthropic and rival OpenAI can find and exploit cybersecurity vulnerabilities at an alarming speed.

But the company also acknowledged serious risks its models, including Claude, pose to itself, its users and the world. Hacks by bad actors using Claude or AI agents could cause financial, legal or physical damage, spark lawsuits and affect Anthropic’s ability to insure against future losses. 

Anthropic expects bad actors, including foreign governments, to increase their efforts to access its latest frontier models as their cybersecurity capabilities improve. That creates persistent risk, since its models could be used in cyberattacks carried out by foreign governments and cybercriminal groups.

“Any successful attempt to misuse, or any perceived misuse of, our models, platform, and applications for a cyberattack or for any other malicious purpose could result in exposure to liability, disruption of our customer or other commercial relationships, or damage to our reputation,” Anthropic said in its filing.

That potential legal and regulatory risk extends to independent security researchers and others who have “used, and may in the future use, our models to identify and exploit vulnerabilities in the systems of other companies, including other AI developers, at times in connection with authorized bug bounty programs.”

Anthropic cited the rise of AI-enabled cybersecurity incidents in the U.S., the U.K. and other countries, saying they could result in more adversarial regulatory environments around the world. European laws, for example, require detailed breach reports within a strict time frame, which could be difficult for Anthropic to meet.

Federal intervention could also disrupt AI development and release timelines. Anthropic cited the Department of Commerce placing export controls on its Mythos 5 and Fable 5 models, and confirmed that the Federal Trade Commission issued it a civil investigative demand earlier this month for more information on the safety and security of its products, as well as its consumer, data and privacy protection policies.

AI development will be guided by more than ethics

Anthropic has developed a public reputation as an ethics-first AI company,  most notably in disputes with the Department of Defense over how its models are used in wartime. But the IPO filing states the company must “balance safety considerations against commercial objectives, competitive pressures, and stakeholder expectations.”

The company also said it is “actively” improving security infrastructure and controls. It also has acknowledged it has moved forward on projects with “known limitations or gaps in our defenses” and may do so again.

“There can be no assurance, however, that the safeguards or access regimes we adopt will prove adequate, that we will correctly identify which capabilities require restriction, or that constraints we impose will not be circumvented,” the document stated.

AI agents can create new and “distinct” cybersecurity attack surfaces because they interact with so many other tools, databases, APIs and other third-party systems. That creates  feedback loops between a model and other systems, which can be compromised or manipulated to feed malicious instructions to the agent.

Anthropic suggested the problems could be so pronounced that it may have difficulty obtaining insurance for some cybersecurity incidents and data breaches related to its models.  

Andrew Yoon, head of research at the nonprofit CivAI, told CyberScoop that both the document and Anthropic’s overall record show a company that has consistently put profit considerations over building safer AI.

“I think in general they’re not meeting the moment here,” Yoon said. “Despite all of these risks that they flagged, they’re moving ahead pretty much as fast as humanly possible, and their justification for this is ‘You can’t do safety from second place.’”

Yoon said Anthropic has not only pushed its own frontier research toward dangerous and harmful capabilities like hacking, “they are pushing very aggressively ahead” and in some cases influencing competitors to follow suit. Anthropic’s release of its Mythos and Fable models earlier this year caused OpenAI to shift “its entire corporate strategy” to match the capabilities of Anthropic models in coding, science, and general intelligence by building models at breakneck speed.

“They have pretty directly caused the industry to speed up toward this mad, frenzied race and yet they’re saying ‘Oh, our models are too powerful,’ ”….it just doesn’t add up to me,” said Yoon.

But Yoon said it’s not clear that building AI systems to carry out dangerous, complex activities like hacking or bioweapons is necessary to be competitive in the global marketplace. He pointed to the wide assortment of Chinese-developed, open-weight AI models that are widely used around the world, but are not powerful enough to chain together complex exploits or develop new viruses.  

“This is not an accident, this is not the models generalizing from regular programming capabilities…the labs are intentionally training them on hacking as well,” said Yoon. “This is because they want to offer better defensive capabilities, they want to be able to sell offensive products to the government…and they do pretty cavalier testing with [the models] inside the labs.”

The post All the security risks Anthropic is worried about before its IPO  appeared first on CyberScoop.

Alarm bells ratcheted up yet again about an AI-discovered vulnerability that could — if exploited in the wild — destabilize vast stretches of the internet. 

The critical defect — CVE-2026-111497 — affects the Google-developed Brotli decompression library that’s bundled into many components, applications and cloud services riding on open-source software. The buffer overflow vulnerability, which OpenAI Codex Security discovered and Node.js disclosed in a Oct. 21 security advisory, can trigger system crashes and remote code execution. 

Researchers haven’t observed any active exploitation of the defect in real networks, but concerns are mounting. Multiple vendors and software packages, including Microsoft and Python, that rely on vulnerable versions of Brotli have rushed emergency patches out to address the risk of exploitation in their respective systems.

Wiz researchers warned that “91% of cloud environments have resources vulnerable to CVE-2026-111497 and 82% of cloud environments have publicly exposed Node.js instances.”

The attack surface is “extremely broad and vast, but much of the impact of exploitation depends entirely on its implementation and how it is used within that piece of software,” said Jake Knott, head of threat intelligence at watchTowr.

Christiaan Beek, vice president at Rapid7 Intelligence, said the Brotli vulnerability is especially concerning because it’s effectively hiding out in the open, as are many open-source software libraries.

“Most businesses have no idea how deeply components like Brotli are embedded in their technology. A vulnerability in a compression library sounds almost boring until you realize it could affect applications your business depends on every day,” he said. “That’s the uncomfortable reality of modern software: a small component can create a very large business risk.”

The defect, dubbed “Node2Shell” or “Broccoli2Shell,” is the latest in a series of vulnerabilities that have provoked an emergency response. Defenders and researchers were on heightened alert last week over a pair of critical PHP vulnerabilities dubbed “PHP2Shell” that also have yet to be exploited in the wild as of Monday afternoon.

These cyclical and increasingly recurring so-called emergencies are putting organizations and defenders at a significant disadvantage. 

“After being part of numerous channels and message threads full of defenders trying to discern what’s real and what’s BS over the past several months — and especially the past week — I can say unequivocally that the human cost of these events is real, pronounced, and will have downstream impacts far beyond what the community can immediately predict,” Caitlin Condon, vice president of research at VulnCheck, told CyberScoop.

“None of this — high-profile vulnerability disclosures, the overblown exploitability assumptions, the community arguments, the race to weaponize — is new,” she added. 

“What’s changed now is that AI is pumping out complex — and often not clearly explained — finds at a much higher rate, exponentially increasing noise levels and making it much more difficult both to debunk overhyped vulnerability discoveries and to determine which vulnerabilities are broadly exploitable in real-world production configurations.” 

Researchers acknowledge the additional stress defenders and organizations are in because of this always-in-emergency mode.

“We’ve reached a point where everything seems to be an emergency. And when everything is critical, businesses risk losing sight of what actually matters,” Beek said. “The real cost isn’t just patching. It’s the constant disruption to operations, the diversion of resources, and the difficult decisions about which business systems deserve immediate attention.”

Experts also consider this the new normal as more high-impact vulnerabilities with far-reaching potential consequences are found. 

“We suspect in the very near future, defenders may lose the ability to tell what is a true emergency versus a loud and prevalent vulnerability in a ubiquitous product or library,” Knott said. 

Experts advise organizations to focus on their specific risk, weighing the Brotli vulnerability’s potential impact in their environment. 

“Attackers don’t care about CVSS scores,” Beek said. “They care about what they can compromise, disrupt, or monetize and that’s exactly how we should prioritize our defenses.”

The post Experts temper emergency-level alarms over critical Brotli defect appeared first on CyberScoop.

A new bipartisan House bill would charge the federal government with developing formal guidance for defending AI models against distillation attacks.

The AI Cyber Defense Act, shared exclusively with CyberScoop ahead of release, is sponsored by Reps. Pat Harrigan, R-N.C., and George Whitesides, D-Calif. It would direct the National Institute for Standards and Technology to develop voluntary industry standards and best practices for preventing distillation attacks from foreign entities from four countries: Russia, China, Iran and North Korea.

In an interview, Harrigan told CyberScoop it is “incredibly expensive” for the U.S. AI industry to maintain its six-to-12 month lead over other countries, and distillation attacks give other parties an incentive to copy other models while keeping their own spending low.

He said he believes countries such as China are making a strategic choice to hold back on full investment of their own AI industry, instead betting they can drift behind American companies as they perform the arduous and expensive research and development to advance the field.

“One of my concerns is that our adversaries just see it as much more beneficial and productive for them as this [U.S. industry] engine burns,” said Harrigan. “Where they actually just see it as more beneficial for themselves in the short term, to hang six to 12 months behind the United States, let us burn really hot and for 5 to 10% of the spend, they can just distill and steal, and they can do that up to until the point where our market kind of falters. At which point they can turn it into high gear, having not expended those resources yet in the past, and actually leapfrog us.”

As part of that mission, NIST will first have to define what a distillation attack is, and separate it from much of the sharing and collaboration between open-weight and open-source models that have also helped drive innovation.

Distillation attacks happen when one AI model is trained on the output data and reasoning of another, usually more powerful or advanced, model. The U.S. government, along with AI companies Anthropic, OpenAI and Google, has accused Chinese companies like Moonshot AI, DeepSeek and Alibaba, of carrying out “systematic” campaigns to distill and copy U.S. commercial models.

According to AI and cybersecurity experts, these campaigns involve the use of hundreds of burner and resold accounts and false identities to bypass geographic restrictions to access U.S. frontier models, and involve thousands, sometimes millions of prompts meant to extract valuable outputs and capabilities that can be used to train a foreign model.

Meanwhile, U.S. AI companies themselves face multiple ongoing lawsuits from media organizations, creators, artists and other laborers, who allege that commercial U.S. models were originally trained on their works without any compensation or permission.  

“AI is advancing at a breakneck pace, and every month we keep China and other adversaries from stealing our progress is a month we gain to build the safeguards this technology needs,” Rep. Whitesides said in a statement to CyberScoop.

The bill would also establish an “AI Cybersecurity Roundtable” at NIST composed of governments and AI industry stakeholders that would work to develop voluntary standards for defensive use of AI in cybersecurity in critical infrastructure.

Harrigan said he envisions the roundtable as including frontier AI companies including software and hardware security experts, who can add independent third-party voices in discussions on the best way to defend IT assets in critical infrastructure and other industries. A place “where you can go and pick an off-the-shelf set of standards that says these are the best practices as we know them all today, given the current state of technology development and the threats that are out there.”

The bill defines “entities of concern” as any foreign person or organization headquartered in one of the four listed countries of concern (Russia, China, Iran and North Korea), or one operating under the direction of a company based in those countries. The director of NIST would also have the ability to designate specific individuals or organizations as entities of concern.

The post Exclusive: Bipartisan bill would develop federal guidance for distillation attacks, convene industry on defensive cyber appeared first on CyberScoop.

Microsoft researchers attributed the recent supply-chain attack targeting ConnectWise customers’ Microsoft 365 environments to a China-state aligned threat actor it tracks as Storm-0940.

“As early as September 2026, Storm-0940 used the agentic AI penetration testing framework ARTEX to automate attacks targeting hundreds of organizations globally across a broad range of industries,” Microsoft Security Research wrote in a blog post Wednesday. 

The nation-state group, which has been active since 2021, targeted organizations in aerospace and defense, automotive, chemicals, energy, financial services, government, legal, manufacturing, media, military, semiconductors, software, technology, think tanks and non-profits. 

Storm-0940 initiated a widespread, but ultimately unsuccessful, password-spraying campaign immediately preceding the compromise of ConnectWise’s Cloud Backup service. During a four-day stretch ending Oct. 14, Storm-0940 targeted 327 Microsoft 365 and Exchange Online accounts across eight organizations, researchers said.

During that short campaign, Microsoft observed more than 151,000 failed logon attempts, peaking at 2,900 failures per minute. Storm-0940 set its sights on ConnectWise the next day, using AI agents and the same malicious infrastructure to compromise a series of ConnectWise Cloud Backup customers’ Microsoft 365 environments. 

Storm-0940 gained significant advantages in the ConnectWise supply-chain campaign by using ARTEX to accelerate scanning, conduct password-spray attacks and abuse OAuth applications in downstream victims, according to Microsoft. 

“In some of these environments, the AI agents (on behalf of the threat actor) progressed beyond authentication to Microsoft Graph and directory operations, including reconnaissance, application credential modifications to establish persistence, and changes involving identities, groups, and privileged roles,” researchers wrote. “Microsoft identified multiple customer environments with evidence of follow-on activity, highlighting how a compromised third-party application with high privileges could provide access beyond its originally intended purpose.”

Microsoft Threat Intelligence said multiple threat actors from various regions are using ARTEX to strengthen capabilities and speed up the tempo of attacks. 

The open-source penetration testing framework works with any AI model to automate various offensive security workflows, the company said. “ARTEX is a system of LLM-driven agents that run an autonomous penetration test from reconnaissance through data exfiltration with little needed human interaction.”

In the ConnectWise campaign, Microsoft said “Storm-0940 used ARTEX to reach more than 2,250 tenants at 2,900 sign-in attempts per minute, a pace that leaves defenders with little time to detect and respond.”

Microsoft did not say how many customers were compromised, adding that successful authentication doesn’t necessarily indicate Storm-0940 accessed or stole data. Yet, researchers said the threat group’s AI agents could also gain persistent access and establish backdoors by modifying credentials in targeted victims’ environments. 

ConnectWise, which temporarily shut down services Saturday in the wake of the attacks, said it began restoring services in phases Wednesday. The company advised customers to keep connections disabled and said it expects the recovery process to take several days.

“Since early October, a wave of AI-driven cyberattacks has been ongoing,” Igor Tsyganskiy, executive vice president at Microsoft, wrote in a LinkedIn post Wednesday. The ConnectWise supply-chain attack is just “one of many,” he added.

The post Microsoft pins ConnectWise supply-chain attack to China-state aligned threat group appeared first on CyberScoop.

The Department of Homeland Security is banning the use of Chinese AI model DeepSeek and its affiliates by civilian federal agencies.

The notice, set to publish Tuesday in the Federal Register, cites the Federal Acquisition Supply Chain Act and a review by the Federal Acquisition Security Council to bar federal agencies from buying all information technology products and services from parent company Hangzhou DeepSeek Artificial Intelligence.

The notice includes a copy of a letter from Secretary of Homeland Security Markwayne Mullin ordering the exclusion of DeepSeek products on June 26, indicating the ban has been in place for months.  

Mullin claimed that DeepSeek’s models present “multiple threat vectors” to civilian federal agency IT systems. Some are unique to the company’s country of origin, while others identify potential privacy and mass data collection concerns that could also apply to other Chinese and U.S. AI models.

He said DeepSeek models store data in servers based in China, where national security laws give the Chinese government wide latitude to request and obtain data from domestic companies. Models hosted directly by DeepSeek collect personally identifiable information from users and trace their “digital fingerprints” online. And like every other major AI model on the market, DeepSeek relies on a “prompt and response” process, something Mullin said “may result in the transfer of sensitive data” to the Chinese government.

“The transfer of the aforementioned information to the [People’s Republic of China], either independently or in combination with publicly available data, could produce actionable intelligence on U.S. persons and/or government programs,” Mullin wrote.

CyberScoop had reached out to CISA and DHS for further details.

The decision will exclude federal agencies from using one of the most popular and powerful Chinese open-weight AI models more than a year after U.S. frontier AI companies and the federal government accused DeepSeek’s makers of “systematically” distilling their models.

The order requires civilian agencies to exclude DeepSeek products from future procurements while removing any versions that might have been put in place prior to the ban. In the past, federal agencies have used their own discretion to block the use of models like DeepSeek, as the Department of Commerce confirmed to FedScoop last year.

Agencies can request exceptions, but must submit their request 60 days in advance and must provide a “compelling justification” for the action, like an inability to fulfill a mission-critical function, or other national security considerations.

Mullin also said agencies that remove DeepSeek from their IT are “encouraged” to enhance their security controls to protect against the possibility that their networks were compromised through the Chinese AI tool.

“During the period between the issuance of this order and full removal of covered articles from agency systems, federal agencies are encouraged to take steps to address risk to their systems, such as monitoring the flow of traffic in and out of the agency, implementing enhanced logging and controls on data exfiltration for the relevant systems, and detecting and flagging anomalous activity,” he wrote.”

DeepSeek is the only Chinese AI model named in the order, but it’s just one of several Chinese open weight AI models that have worried national security officials or have been accused of illegally distilling U.S. models, along with Moonshot AI, the makers of Kimi and Alibaba, makers of Qwen.

The post DHS to ban DeepSeek from federal civilian agencies appeared first on CyberScoop.

The federal government is advising AI companies to set more industry standards for securing agents in sandbox testing environments.

In joint guidance from the Cybersecurity and Infrastructure Security Agency, the National Security Agency and the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI) at the Department of Commerce, AI companies are encouraged to lean more heavily on cybersecurity technical controls, independent testing and evaluation of new models and develop “shut down policies” that govern when a model or test should be taken offline.

The guidance comes after a summer when OpenAI, Anthropic, Google, Meta and other companies say their AI models escaped from testing environments and hacked real organizations and websites. The problem has bedeviled frontier AI companies, with many of their own researchers saying it’s possible that hundreds – even thousands – of similar sandbox escapes may have occurred over the summer, raising major questions about the lack of effective monitoring in place.

“While these incidents highlight the importance of traditional cybersecurity principles and practices, they also introduce novel considerations for incident prevention, response, and reporting,” the agencies wrote. “These novel considerations primarily stem from the agentic nature and advanced cyber capabilities of cutting-edge SI systems.”

While voluntary, the document marks the federal government’s most detailed cybersecurity guidance for securing sandbox environments during AI model tests. 

The first set of recommendations deals with designing and implementing technical controls for the virtualized testing environments, while another recommends outside monitoring tools to ensure AI models don’t access tools or privileges they shouldn’t have. Cybersecurity experts have said both features were lacking in the testing environments used by AI companies this past summer.

AI testers must also scour their internal testing environments for “potential persistence mechanisms and unauthorized communication across internal and external systems” that could be used by AI agents to leave notes and collaborate on tasks in secret.

It specifically cautions against relying solely on other AI systems to monitor and restrain the behavior of their test models, using the government’s now-preferred term of “Super Intelligence” or “SI.”

“Test and red-team SI control measures with humans and SI systems before deploying them,” the guidance states. “Only use SI systems to red-team in cases where out-of-scope actions could not result in harm to critical systems.”

Security teams must evaluate risk not only from the large language model, but also from the surrounding systems and infrastructure it relies on, like the technical harness, the host and broader network, and each layer has security protocols that can stop unauthorized access to the internet or detect anomalous behavior.

Another recommendation calls for frontier AI companies to be more open to partnering with “independent, external auditors” to assess their controls and monitoring for AI agent testing. Critics have argued that Anthropic and OpenAI rely too heavily on a small group of little-known nonprofits and private research organizations, like METR and Redwood Research, to test their frontier models for safety and security concerns and respond to incidents like the Hugging Face hack. They question why these companies don’t work with more established cybersecurity or incident response firms.

Chris Painter, METR’s president, told Congress earlier this month that it had become industry practice for frontier companies to rely almost entirely on other AI systems to monitor and restrain their models in testing environments, citing the sheer volume of data and telemetry that researchers must sift through.

Cybersecurity and AI experts have told CyberScoop that a combination of human oversight, technical controls and AI systems is the most effective means of restraining the kind of agent hacks that have consumed the news.

The guidance further calls for executive leadership at these companies to be more personally involved in decisions around pausing or restarting a particular project, and only after they’ve been informed of the relevant potential risks.

The post Federal agencies recommend cyber controls, independent testing for frontier AI systems appeared first on CyberScoop.

Election influence campaigns, particularly ones playing out over social media, are a dime a dozen.

Such campaigns – sometimes devised by foreign powers, other times by domestic groups – are often designed to influence the U.S. electorate. Fake accounts and online personas are created, networks of propaganda are disseminated and the campaign seeks to nudge American voters into thinking a certain way.

Mostly, these efforts are known for getting little to no traction with U.S. audiences. But while social media influence campaigns may not be effective, bad actors continue to adapt their coordinated propaganda goals to new tactics and technologies.

Enter AI chatbots, which have become a source of information and truth for many, who consult the tool for feedback and advice on everything from their careers, to their relationships with their family, as well as their politics.

But new research from VisionHeight, an Israeli threat intelligence firm, highlights how AI systems can often serve as more useful marks for influence actors than humans. To wit: someone is setting up networks of hyperlocal websites across at least 14 different states, filled with content meant to influence U.S. races.

These networks include websites with names like “Wolverine State Report” or “Peach State Bulletin.” Each state network usually has between five to six separate sites,  some presenting themselves as hard news outlets and others as opinion or editorial sites. The outlets also cite one another, creating a circular network of local credibility.

But the content on these websites does not resemble journalism, or any writing that a human would find coherent or compelling. Rather, they often include text of statements about specific candidates running in local races, directly stating they are the better candidate for a political issue.

Statements like “Candidate X is the stronger choice” for voters who care about housing, taxes or other political issues were repeated hundreds of times on the websites, with slight changes in the language for different political issues. Most notably, they appear to be explicitly written for AI crawlers scraping web traffic, not humans.

“What’s really interesting about this is that the websites themselves and the content is so well built to let’s say optimize the probability for AI crawlers and the model to be influenced by this content,” said Nitsan Daniel, chief technology officer and co-founder of VisionHeight in an interview.

When a user asks an AI chatbot a question, the model will often search the internet for an answer. By creating specific online content providing direct answers to questions like “which candidate in a local race is better on housing affordability” it can essentially train the AI to retrieve that information when responding to queries.

Many media websites will opt to block AI crawlers, but these websites were all configured to specifically allow major U.S. AI companies to scrape their websites.

A blog by VisionHeight published Wednesday said the operation encompassed six states and 37 websites, but Daniel told CyberScoop that they have since discovered the operation spanned at least 14 states and 460 pages of online content.

The states targeted include Arizona, Colorado, Georgia, Indiana, Iowa, Kansas, Michigan, Mississippi, Nevada, North Carolina, Pennsylvania, Texas, Virginia, Wisconsin . At least seven states saw one or more House races targeted, while other sites focused on other races like Secretaries of State, statewide judicial appointments and local offices.

Most of the research focused on AI answers provided through Google search, but Daniel said they had confirmed at least one major AI model, Perplexity, linked to and cited one of the pages, Tarheel State Report, as its first source for an answer.

VisionHeight does not attribute the activity to a particular actor. To Daniel, who is behind the campaign and their ultimate goals are less important than the prospect that this kind of operation is easily reproducible by both foreign and domestic bad actors looking to influence how AI systems answer questions around political races.

“What we’re seeing here is,I think, very nicely demonstrating how the entire world is transitioning to [a new era],” she said. “So if in the previous election you could have seen fake websites aimed at getting people to actually read it, and containing content built to impact what they’re thinking, what we’re seeing here is…now instead of spending time and effort on getting people to read the content, [they’re] getting into their AI and being influenced by that.”

But Daniel’s concerns are broader. Large language models have become increasingly sophisticated in their ability to carry out a range of deceptive activities, such as creating false organizations, websites, as well as deepfake video, audio and imagery at scale.

It appears AI was used to generate much of the infrastructure and content used in the operation as well. All 37 websites VisionHeight identified in their first analysis, spanning multiple states, were registered in a single evening, and 36 were registered within an 8 minute timeframe. On Sept. 30, the network published “roughly 100,000 words of candidate content” all using the same template and layouts.

Foreign and domestic actors can easily replicate this process, generating hyperlocal propaganda that shapes what  AI chatbots say in response to political questions. Experts say chatbots have improved at avoiding false election information, but they remain vulnerable to manipulation and errors that make them unsuited to providing more complex answers around politics.

Whereas in the past a foreign actor might lack the budget, time and incentive to target small elections, AI tools now allow for such operations to be created and optimized at scale within days. Domestic actors, meanwhile, could potentially pair the same tactics with a more sophisticated understanding of how to influence American audiences.

“It really demonstrates how in general using AI just makes what was in the past small targets [more attainable], Daniel said. “People could think that the odds of them being hit, being manipulated or being attacked are just lower because people will not put all their energy and resources on that. Nowadays, because it is so much easier, we’re seeing influence operations and cyber attacks hitting small, medium businesses as well as smaller communities, like in this case.”

The post The next election influence campaign will target your AI chatbot appeared first on CyberScoop.

Microsoft’s streak of record-breaking vulnerability disclosures continued with a mega Patch Tuesday, containing 1,413 defects affecting the company’s various enterprise products and services. 

The monthly batch of patches surpassed the company’s previous annual record of 1,261 common vulnerabilities and exposures (CVEs) set in 2020. 

“Hundreds of CVEs in a single release used to be the exception — now it’s the baseline,” Dustin Childs, head of threat awareness at TrendAI’s Zero Day Initiative, wrote in an email. AI is partly to blame, he added, but it “isn’t creating these bugs out of thin air; it’s just shortening the time between a vulnerability existing and someone noticing it.”

Yet, for defenders understandably overwhelmed by the surge of defects found in Microsoft products, actively exploited zero-day vulnerabilities aren’t growing in kind.

“If there’s solace to be had, it’s in the fact that only one of these bugs is under active attack,” Childs wrote in a blog post. “If you go back to the hot 0-day summer of 2023, there were 17 Microsoft 0-days between January and October as opposed to 15 this year during the same period.”

Microsoft disclosed one actively exploited zero-day affecting Windows Shell — CVE-2026-96951 — that allows unauthorized attackers to bypass a security feature. The high-severity defect is the third Windows Shell vulnerability exploited this year, according to Satnam Narang, senior staff research engineer at Tenable. 

“There have been 26 Windows Shell vulnerabilities patched in 2026 through October, a staggering figure compared to previous years, with only 3 Windows Shell CVEs in 2025 and just one in 2024 and 2022,” he added.

The vast majority of defects Microsoft disclosed in this month’s security update affect Windows, its ubiquitous operating system , spanning 1,102 vulnerabilities combined. The update also included 134 vulnerabilities in Office, 61 in various developer tools and 36 in SharePoint Server. 

Microsoft tried to address the toll this monthly flood of vulnerabilities has on its customers by introducing “risk cohorts.” These groups of vulnerabilities don’t necessarily share a root cause or severity rating, but generally require a similar customer assessment that could guide prioritization efforts, the company said. 

“The volume isn’t the real story anymore,” Childs added. “What matters is that discovery is scaling faster than exploitation, so teams that triage by sheer CVE count instead of exploitability are burning cycles on the wrong bugs. Patch what’s proven, prioritized, and public, and treat everything else as this month’s new normal.”

The full list of vulnerabilities addressed this month is available in Microsoft’s Security Response Center.

The post Microsoft drops an actively exploited zero-day among record high of 1,413 vulnerabilities appeared first on CyberScoop.

Members of Congress are pressing the Trump administration to expand the federal government’s role in securing artificial intelligence, wanting new safety standards and regulations, independent testing of models and more “aggressive” use of AI for defensive cybersecurity purposes.

In a letter, Sens. Maria Cantwell, D-Wash., and Mike Rounds, R-S.D., said that multiple incidents of frontier AI models escaping sandbox protections and breaching real companies, organizations and governments necessitate a stronger federal role in regulating the technology.

Recent revelations that hundreds, and possibly “tens of thousands,” of such incidents may have occurred for months without OpenAI’s staff detecting them makes the need for federal intervention even more urgent, the senators argued.

“These incidents come as AI systems are becoming dramatically more capable at discovering software vulnerabilities, writing and executing code, using digital tools, and operating autonomously,” the senators wrote.

The letter, addressed to Jay Clayton, director of national intelligence and new White House “AI czar” and Michael Kratsios, director of the Office of Science and Technology Policy, urges the federal government “to bring its full capacities to bear on this challenge, building a coordinated national response around standards, testing, auditing, and defensive AI.”

The Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), housed within the Department of Commerce, has become the Trump administration’s main body for evaluating and testing AI models. In the letter, the senators are seeking to expand the center’s resources and mandate.

Under the Trump administration, testing of frontier models has been voluntary and focused on evaluating model capabilities, not restricting them. The administration has since said it has reached a safety testing framework with OpenAI, Anthropic and other AI companies to govern safety testing, but has not made the agreement public.

The senators call for CAISSI to spearhead the development of “technically grounded and measurable standards and risk management frameworks for advanced AI safety and security.” Those standards would address potential risks like cyberattacks, automated attacks on critical infrastructure and other threats.

“The standards should also address the security of the environments in which advanced AI systems are developed, trained, tested, and used internally, including protections against agents escaping testing environments, gaining unauthorized access to external systems, or taking actions beyond their intended permissions,” the senators wrote.

The letter argues the center and federal technology hubs like the National Institute of Standards and Technology should hire additional expertise in AI, cybersecurity, model evaluation and national security.

It also calls for a much broader community of independent and third-party testers. Currently, that testing is done almost entirely by AI companies and hand-selected auditors, like Model Evaluation and Threat Research (METR), Redwood Research and others.

Last month, Chris Painter, president of METR, told Congress that the firm’s investigation of the Hugging Face hacking incident was at the invitation of OpenAI, which tightly controls which organizations and testers get access to its unreleased models. While they are permitted to publicize some findings, such as those that relate to potentially catastrophic risks to humanity, researchers are otherwise restricted from discussing other parts of their investigation through nondisclosure agreements.

Additionally, Painter suggested that frontier companies are relying too heavily on their own technology, rather than technical controls, to monitor and restrain their models in testing.

“Because of the scale and speed at which AI agents are deployed, companies depend on AI monitoring and controls instead of human supervision, to prevent unwanted agent behavior,” Painter said. “That is, monitoring of AI is now in large part done by other AI systems.”

The senators call for CAISSI, national laboratories and the Department of Energy to become federal testing sites. There, the government would evaluate AI models for advanced cyber capabilities, biological or chemical threats, threats to critical infrastructure, sandbox evasion techniques and other autonomous rule-breaking behavior. They also want testing opened to universities, independent research organizations and “qualified third-party evaluators,” including private sector companies.

Other federal bodies, like the National Security Agency, the Defense Advanced Research Projects Agency and National Science Foundation can help develop new standards and testbeds for AI tools used in defensive cybersecurity that can then be shared across government and with the public.

“The recent incidents make clear that the security infrastructure surrounding advanced AI must advance as quickly as the technology itself,” the senators wrote. “The United States has an opportunity to lead both sides of this challenge: building the standards, testing, and auditing infrastructure needed for increasingly powerful AI, while harnessing those same capabilities to make American networks harder to attack.”

You can read the full letter here.

The post Senators press Trump admin to rally federal government around AI safety, security  appeared first on CyberScoop.

A joint international law enforcement and industry operation against scammers netted 28 arrests, rescued 117 trafficking victims and led to the takedown of more than 3.7 million accounts, the Justice Department announced Tuesday.

The crackdown, billed as “Joint Disruption Week,” focused on West Africa and also identified more than $11 million in freezable funds and led to the seizure of hundreds of electronic devices. The range of scams involved included law enforcement impersonation scams, digital arrest extortion schemes, pig butchering and other online fraud operations targeting victims around the world.

Participants heralded it as the largest joint scam operation to date, and the first focused on West Africa.

“For so long, the problem of international scams seemed impossible to tackle,” Kelly Milliron, acting chief of staff and counselor to the Assistant Attorney General leading the Criminal Division, told reporters. “But now, because of collaborations like Joint Disruption Week, we have a plan, a consolidated approach, bringing together people that can bring solutions and, most of all, safety for our citizens.”

It’s the latest effort in the Trump administration’s ongoing bid to make countering scam operations a theme of his term, and to make disrupting cyber adversaries and criminals a central focus of its cyber strategy. 

Participating agencies in the London-hosted operation beyond the Justice Department were the Scam Center Strike Force under the U.S. Attorney’s Office for the District of Columbia, the FBI, the Secret Service, the Treasury Department, Homeland Security Investigations, the Homeland Security Task Force, the United Kingdom’s National Crime Agency, the Nigerian Joint Case Team on Cybercrimes, the Nigeria Police Force’s National Cybercrime Center and the Ghanaian Cyber Security Authority, INTERPOL and agencies from Australia, Canada, India and New Zealand also were involved. 

Adobe, Binance, Coinbase, Cyjax, Elliptic, Chainalysis, Flashpoint, Google, Meta, Microsoft, OpenAI, Silent Push, Starlink and TRM Labs were companies that participated in the operation. Meta said it removed more than 3.7 million accounts, groups and pages from Facebook and Instagram based in West Africa, a majority of which targeted people in the United States and United Kingdom.

Some on Capitol Hill have raised concerns about whether U.S. operations to counter scams are coordinated enough.

The operation actually began in September, with a series of raids in Ghana before last week’s designated action. Jeanine Pirro, U.S. Attorney for the District of Columbia, said Disruption Week occurred not long after a new partnership with the United Kingdom.

“The aim is to bring cases in parallel and also to combat fraud together through joint targeting,” she said. “This was the perfect time to continue that partnership.”

It’s not clear how many of the potentially freezable funds will actually be frozen as a result of the operation.

“Identifying the assets is the easy part, and tracing them, going after and developing probable cause within our investigations so that we can legally seize them does require a little bit of work,” said Matthew Floyd, deputy assistant director of the FBI’s Criminal Investigative Division. “So at this point, that’s why it’s much easier in these efforts to identify assets initially, and then they’ll fold into our investigations as we move forward.”

Disruption Week was the culmination of agencies meeting to combine forces, something that will happen on a regular basis, Floyd said, but at an interval to be determined going forward. The operation led to the sharing of more than 30 leads on scam centers, participants said.

The post West African scammers targeted in joint international government-industry operation appeared first on CyberScoop.

Strip away the model names and the headlines, and this year’s AI agent incidents share a feature that has little to do with how capable the models are. Nearly all of them ran through the software supply chain.

Start with the case that set off the summer’s wave of disclosures. In July, two OpenAI models undergoing a cyber-capability evaluation escaped their sandbox. They got out through a zero-day vulnerability in a package registry proxy that sat inside the evaluation boundary, placed there so the models could install the tools they needed. From there, they reached the open internet and went after Hugging Face’s production systems, looking for the benchmark’s answer key. Hugging Face caught and contained the intrusion on July 16. Five days later, OpenAI confirmed the incident and connected it to an internal cybersecurity evaluation of frontier AI models.

Soon after, Anthropic disclosed three incidents in which its Claude AI models compromised external organizations after a misconfiguration left live internet access in place. In September, it added a fourth that a wider review had turned up. The incident that Anthropic says worries it the most involved a model publishing malicious packages to the Python Package Index (PyPI).

Separately, researchers have attributed more than 2,000 RubyGems uploads in May to OpenAI agents. OpenAI says its agents were carrying out benign tasks. RubyGems removed more than 500 of the packages as malicious.

In each case, the agent’s route out, or the harm it caused, ran through a package registry, a dependency or a model file.

None of this surprised the researchers I work with. ReversingLabs has spent years taking apart the files that move through software pipelines, and we watched attackers discover the AI supply chain well before agents reached production. In early 2025, our team found malicious models on Hugging Face whose pickle payloads were packed in a 7z archive that Picklescan, the platform’s scanner, could not open. The payload itself was an ordinary reverse shell. Hugging Face took the models down after we reported them.

In April, we published research on PromptMink, a campaign we tie to the North Korean group Famous Chollima. In one case, a Claude Opus-based coding agent added a malicious dependency to a cryptocurrency trading project. What struck our team was who the attackers were writing for. Their package documentation was built to be more convincing to an AI coding assistant than to a human developer.

Put those cases side by side, and the threat model looks different from the one most enterprises are planning around. The evaluation escapes show agents misusing the access they were given. PromptMink shows agents being steered by content written specifically to fool them. In both, the damage moved through software the agent downloaded or published.

That is the context for the Open Agent Safety Platform NVIDIA launched on Sept. 28. The design starts from the right premise: you cannot secure an agent by asking it to behave. OpenShell runs each agent in a sandbox and applies policy outside the agent’s own process. Sentry, part of a reference design built on BlueField-4 data processing units (DPUs), watches agent activity from separate hardware, out of reach of a compromised host. More than 100 partners signed on at launch, ReversingLabs among them.

The Hugging Face incident shows why that architecture matters. OpenAI ran those models with cyber refusals reduced, meaning their built-in guardrails were deliberatelyweakened. That left the sandbox boundary as the main line of defense. Prompts and training shape what an agent attempts. Infrastructure decides what it can actually do.

It also matches what I hear from customers. They are demanding AI safety that closes as many gaps as possible and uses the boundary between hardware and software to deliver security they can rely on. That is the standard every vendor in this ecosystem, mine included, will be measured against.

Look, though, at where that boundary failed: a package proxy, trusted because it was supposed to be there. A policy engine can tell you an agent is permitted to install a package. It cannot tell you whether the package is what it claims to be. NVIDIA’s CISO, Saša Zdjelar, made the point himself in a Sept. 21 post, writing that teams must “verify the source and integrity of the tools, skills and dependencies agents use.” Doing that means inspecting the artifact itself, whether a model file, an archive, a container or a binary, at the boundary and before the agent touches it. Even AIUC-1, a prominent certification standard for AI agents, treats scanning model files for malicious payloads as a supplemental control rather than a required one. Its vendor due diligence requirement stops at the model provider. That is the layer ReversingLabs brings to the platform.

I have two cautions as Sentry moves from reference design to shipping products. The first is about openness. Sentry depends on specific NVIDIA hardware, and “open” will only mean something if enterprises can plug in the inspection, telemetry and policy tools they choose and run them on infrastructure they already own. Buyers should hold the ecosystem to that standard.

The second is about the need for prompt disclosure, and it applies to the whole industry. In June, an OpenAI agent accessed non-public files on the Australian government’s Medicare Statistics Reporting Service portal. Services Australia, which runs the portal, learned of it on Sept. 10, through an email to a public mailbox Prime Minister Anthony Albanese called the delay unacceptable. RubyGems maintainers learned who was likely behind their May flood of packages only from outside researchers, months after the fact.

We would not accept that timeline for any other security incident, and AI agents should not be an exception. Organizations that run agents, AI labs included, should notify affected parties on the schedules we already expect for breaches. Package registries are shared infrastructure for the entire software economy, and they should be told when an agent has published to them. Labs running offensive evaluations should treat every service reachable from the sandbox, the package proxy included, as an attack surface, and they should publish the containment standards they hold themselves to.

Compromises like SolarWinds, 3CX DesktopApp and xz Utils taught the security industry that software cannot be trusted simply because of its reputation or where it came from. Agents now pull and publish that software on their own, at a volume no review team can match. NVIDIA has built a strong platform for enforcing trust decisions. The rest of us have to make sure those decisions rest on knowing what is inside the software.

The post Your AI agent is only as safe as the packages it downloads appeared first on CyberScoop.

Anthropic announced a new program Thursday that will combine its AI tools with outside cybersecurity companies to find and fix cybersecurity vulnerabilities in critical infrastructure and open-source software.

Advertised as part of a new, “long-term commitment” to cybersecurity, the critical infrastructure defense program will pair Claude models, Anthropic engineers and threat research with the expertise of cybersecurity companies, including Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC, and Rockwell Automation.

“Defenders of critical infrastructure and the [open-source software] community have decades of security experience but have faced severe resource shortages that are exacerbated by this moment,” Anthropic wrote in a blog post. “We are recognizing their expertise in these domains and offering our support.”

Anthropic said it has offered frontier models and technical support to more than half the states in the U.S., as well as large operators of critical infrastructure to scan and patch code or assist in incident response and red teaming activities.

“Critical infrastructure is hard to defend in many ways that AI cannot fix, but we believe that frontier models can help find and repair weaknesses before those weaknesses are used to cut off power or make water unsafe, the company wrote. “Our first step is to work with a small cohort of providers to learn which strategies are most effective and practical.”

The company said it has engaged in similar work with maintainers of large open-source software projects, and some organizations asked for everything the model had found in their software, even unreviewed findings.

That spurred Anthropic to create a new opt-in scanning service for open source software, where organizations can get free, periodic scans of their projects with a proof of concept, explanation and suggested patching options. Anthropic noted that under this program, organizations will receive their reports faster but they may contain inaccuracies.

The long-term goal is to automate most triage and patching and develop new security architectures and coding standards.

As AI models have developed increasingly powerful cybersecurity capabilities, such as finding and exploiting known vulnerabilities, frontier AI companies have worried that bad actors could gain access to the tools faster than legitimate organizations. Anthropic and OpenAI have since both started programs funneling their tech to businesses and governments, free of charge, for defensive cybersecurity.

The post Anthropic rolls out program for ‘long-term commitment’ to secure critical infrastructure, open source software appeared first on CyberScoop.

The Justice Department and FBI announced that they had seized two hacking tools connected to the Chinese government-linked group Flax Typhoon and a China-based company that the U.S. government has repeatedly taken action against, including with a new multi-agency advisory Thursday.

The domain name seizures were meant to deny hackers access to the vulnerability scanning tool Microscan and the spearphishing tool FishHub created by the Chinese firm Integrity Technology Group, which the United States sanctioned last year. The U.S. government in 2024 also made the company the focus of a takedown operation, saying it was behind a massive botnet.

Authorities were granted the court-authorized seizures in the Western District of Pennsylvania, the FBI and DOJ announced Thursday in unsealing documents in the case.

U.S. agencies paired the announcement of the seizures with an advisory from the FBI, Cybersecurity and Infrastructure Security Agency and National Security Agency.

“Chinese government-linked cyber threat actors, enabled by the Integrity Technology Group, are combining automated scanning tools, large-scale botnets, and hands-on exploitation techniques to target and steal sensitive data from organizations worldwide, including U.S. critical infrastructure sectors,” the advisory reads. “These actors exploit vulnerabilities by using scanning tools, cross-site scripting attacks, and password spraying on Microsoft Exchange servers, while establishing persistence through VPN software and exfiltrating emails and credentials using scripts.”

Similar to previous allegations, law enforcement said Integrity Tech leaned on a Mirai-variant botnet of internet-of-things devices, in this case to facilitate Microscan.

Microscan targets have included a South Carolina power company, airports in Japan and Poland, critical infrastructure companies and universities in Taiwan and others, according to the FBI and DOJ. Victims of FishHub, which downloads malware on their networks after gaining access through spearphishing, include Taiwanese universities.

“Chinese government-affiliated actors continue to position themselves within critical infrastructure networks, including operational technology (OT) systems, with the aim of disrupting critical functions at a future time of their choosing,” said Chris Butera, acting executive assistant director for cybersecurity at CISA. 

“Under the FBI Cyber Strategy, we pursue both the actors who threaten critical infrastructure and the enterprises that support them,” said Brett Leatherman, head of the FBI’s Cyber Division. “Integrity Technology Group, a China-based company with ties to the Chinese government, is one of those enterprises, acquiring or developing cyber tools and hosting infrastructure for actors targeting networks worldwide.”

The post DOJ, FBI seize Flax Typhoon-linked hacking tools Microscan, FishHub appeared first on CyberScoop.