The UK’s AI Security Institute said that their AI research system took “unsanctioned” actions over the internet to engage in “sustained, potentially harmful activity directed at real people.”

According to a blog post and technical report released Tuesday, the organization noticed “unusual data transfers” leaving their research systems through the pseudo-anonymous Tor network on July 28. The models, which included Anthropic’s Mythos 5 and OpenAI’s GPT-5.6-Sol models, were being tested for their cybersecurity capabilities.

Like the OpenAI and Anthropic incidents before it, the offending agents were attempting to solve a challenge. On 10 of 122 runs taken by two models, they took a combined 19 malicious actions, including attempting to insert malicious code into a real open-source software project and creating fake online identities to contact human maintainers and press them to approve it.

It also inserted malicious prompt injection instructions in places “where it reasoned that other automated AI systems might pick them up and execute them.” Sometimes multiple agents would collaborate on a task, with one model leaving public messages or notes for the other on GitHub.

However, AISI said their incident differs in one critical way.

“Importantly, this was not a case of a model escaping its secure test environment, or ‘sandbox,’” the institute wrote in an unsigned blog. “As was standard in our cyber testing, we had intentionally permitted internet access, and model-provider cyber classifiers were deliberately disabled – conditions that do not reflect how frontier models are made available to the public.”

The institute said that while their own design and evaluation choices “to some degree” enabled the malicious behavior, the models still displayed “novel, potentially deceptive behaviors” and were executed “to an extent and severity that we did not anticipate.”

The same day, OpenAI acknowledged AISI and another third-party cybersecurity tester,  private firm Irregular, had recently reported incidents where their models exceeded “beyond their intended testing boundaries.”

That included GPT-5.6-Sol reusing GitHub tokens left by another agent to attempt to recover accounts and to access a DNS server that contained malicious payloads meant to exploit a software vulnerability.

The setup did not work, there is no evidence any real resolver queried it, and the infrastructure was removed when the evaluation ended,” the company wrote in an unsigned blog post Tuesday.

OpenAI said it would review its own third-party testing procedures to focus on higher risk evaluations, assess requests by third-parties to enable internet access, stop conditions and other features for their models.

The incident with Irregular occurred on July 29 during a Capture-the-Flag cybersecurity evaluation of OpenAI’s models. Due to a “misconfiguration” the models were allowed to access the public internet, where they encountered and exploited a real domain, mistakenly believing it was still in a test environment.

OpenAI said an investigation by Irregular is ongoing, but also found that the models had found and used credentials for the site at one point. The blog also references other additional potential cybersecurity incidents.

“Irregular has informed us that all of the issues identified pertaining to the incident are no longer active and relevant safeguards were added to the testing environment,” the blog said. “Irregular has also communicated about related incidents involving other labs from the same testing environment.”

CyberScoop has reached out to Irregular for comment.

The incidents were made public the same day that the White House met with Anthropic, Open AI and other frontier AI companies to preview a new framework for evaluating models before they’re released publicly. Some media outlets have reported that after an executive order, export controls and other actions, the administration does not plan to make the new framework public.

The post AISI, OpenAI report more ‘unsanctioned’ model hacks appeared first on CyberScoop.

The Trump administration’s haphazard and opaque interventions into artificial intelligence security matters could catapult Chinese alternatives into broader acceptance, posing new security risks altogether, a group of Democratic senators wrote to top administration officials Monday.

The five senators said that the administration’s handling has alternated between too passive, such as when OpenAI models escaped testing in the Hugging Face hack last month, and overstepping, such as when the Commerce Department suspended access for any foreign national to Anthropic’s Fable 5 and Mythos 5 in June.

“The Administration’s ad hoc and unpredictable approach undermines U.S. competitiveness, heightening market incentives to adopt open weight models from vendors based in the People’s Republic of China (PRC),” wrote Sens. Kristen Gillibrand of New York, Adam Schiff of California, Mark Warner of Virginia, Chris Coons of Delaware and Mark Kelly of Arizona.

In the Hugging Face hack, the senators wrote that “the Federal Government cannot be passive as these capabilities emerge.”

In the case of the Fable 5 and Mythos 5 suspensions, the senators said that the administration “utilized an infrequently used authority to direct Anthropic to suspend all access to its Fable 5 and Mythos 5 models for foreign nationals (including foreign national employees inside the United States) citing an undisclosed national security concern later described as a narrow jailbreak finding.”

Because Anthropic couldn’t immediately assess users’ nationality, the firm had to disable both models for everyone. The administration and Anthropic negotiated for 18 days behind closed doors before reaching an agreement, the lawmakers complained.

“While the Administration may have been responding to real security concerns to protect the United States, even justifiable interventions can create broader harm if the standards and decision-making processes are opaque, ad hoc, or unpredictable,” they said in their letter to leaders in the White House, Office of the National Cyber Director and departments of State, Treasury and Commerce. “Moreover, when the Executive Branch exercises authority delegated from Congress, such as in the conduct of export control administration, it is essential that it keep Congress fully apprised of its actions and procedures.”

During the time Anthropic was under export controls, the stock price of “an entity-listed Chinese lab” nearly doubled, the senators said. And while Hugging Face was breached, the company “had to” rely on a Chinese open-weight model due to guardrails on U.S. frontier models.

“If American models are perceived as subject to sudden access disruptions based on a black-box U.S. Government process, or as unreliable because U.S. AI labs are overcorrecting in the face of this black-box process, companies and governments in the United States and abroad may hedge by adopting Chinese or other foreign models instead,” the senators contended. “That outcome would undermine U.S. technological leadership while increasing exposure to systems that may carry risks of PRC or otherwise directed censorship, espionage, IP theft, and other supply chain security risks.”

Their letter asked for answers to questions about the standards the administration uses to determine the national security risks a frontier model presents, what legal authorities it will use to invoke restrictions, which agencies are responsible for which decisions and more.

None of the offices or departments the letter was addressed to immediately responded to a request for comment.

The letter follows inquiries at the state level, where 15 attorneys general asked OpenAI for more information regarding the security incident at Hugging Face.

The post Dem senators criticize Trump administration decisionmaking on AI security risks appeared first on CyberScoop.

Frontier AI companies often refer to their models as “helpful assistants” or try to compare them to entry-level employees.  

But new research from the UK’s AI Security Institute reinforces how large language models suffer from a common flaw that would land many human employees in hot water with their employers: they cheat.

In other words, these models are so committed to completing their tasks that they will break the rules, cut corners and deceive  their own users to accomplish them.

“Every model we have tested for this behavior attempted to cheat,” the AISI report said.
“Models did not reliably report this behavior when asked, and often did not reason about it in their chain-of-thought, suggesting that detecting cheating will likely require robust monitoring methods.”

The research tested OpenAI’s ChatGPT 5.4, 5.5 and 5.6 models, along with Anthropic’s Claude Opus 4.7 and Mythos Preview.

The AISI ran models through a series of “Capture-the-Flag” cyber evaluations, where they obtained points by performing offensive cybersecurity related tasks, like exploiting a particular vulnerability or reverse engineering compiled code.

The research defines “cheating” as “taking an action that is out of scope for the task or explicitly disallowed by the rules, in order to achieve a goal through a shortcut, workaround, or unintended solution that the task was not meant to, or should not, permit.”

It captures a range of observed behaviors from models, like searching the internet for solutions, attacking or escalating privileges on unrelated systems, probing evaluation software to gain access to the task solution.

Models also failed to acknowledge when they were cheating to complete a task, and less than 50 percent said the rule-breaking was “wrong” when challenged on it by a user.

A model’s propensity for cheating was not related to its capability, meaning newer or advanced models aren’t more likely to cheat. Rather, researchers think it stems from the techniques used during a model’s training and alignment that are to blame.

But even if rates of cheating remain constant, the problem could still worsen over time. As newer models in the future could become more proficient and learn more effective cheating techniques.

This deception also makes it difficult for labs like AISI to verify their own work, which relies on evaluating trustworthy outputs from AI systems.

The research underscores how AI systems can go to drastic lengths to complete their task, including blowing through or circumventing a company’s IT and cybersecurity protections.

In one instance, AISI researchers said a model was inadvertently given a cyber capability evaluation that was misconfigured and impossible to solve.

“The model tested was so persistent in attempting to cheat that it wrote and ran code on an external service, hosted on the open internet outside of AISI’s systems, in an attempt to access our evaluation infrastructure, triggering a security alert in AISI’s systems,” the report said.

While AISI said there were no data leaks or damage from the incident, the model could have successfully accessed their evaluation system had they not had monitoring in place. The institute said it implemented further controls on internal systems in response to the test.

The researchers said there are “significant consequences” to a status quo where we can’t trust models not to cheat. The behaviors are especially problematic in areas like AI safety and security research, as well as cyber operations and military decision-making, where trust outputs from the AI systems are critical.

Today, AISI said it can detect LLM cheating through a mix of manual review and LLM monitoring, but that may not always be true, and future models may be better at hiding their actions from human overseers.

“A more fundamental fix would be to train the models not to cheat in the first place – but given this kind of behavior was reported in frontier models more than a year ago, robustly aligning it away may not be easy,” researchers wrote.

The post AI models keep getting caught cheating appeared first on CyberScoop.

After a year and a half spent downplaying calls for AI safety regulations, the Trump administration has sharply reversed course, embracing a level of government scrutiny of frontier AI systems before public release–a far stricter stance than the Biden administration took.

An executive order designed to be friendly to the AI industry was meant to let the federal government briefly review some new models on a voluntary basis.

When the Trump administration, suddenly and without much warning, slapped export controls on Anthropic’s Fable 5 and Mythos 5 in response to private sector threat intelligence reporting, the U.S. AI industry officially entered its regulatory era.

But key questions and gaps remain. It’s not clear why the administration drew the line where it did, or whether they will move it again in the future.

While newer models like Mythos and OpenAI’s Daybreak do have stronger cybersecurity capabilities, the private sector reports the administration relied on describe capabilities already available in older commercial, open-source and Chinese models that nearly anyone can access.

CyberScoop spoke with current users of the latest frontier models, including OpenAI’s ChatGPT 5.5 and Fable 5, to learn more about what these models are currently capable of in offensive and defensive cybersecurity.

Cybersecurity experts and former government officials say the administration may be playing catch up on threats that have been building for years as it has more fully realized the national security implications of the technology.

Are the models breaking new ground or just breaking things? 

Users of Chat GPT 5.5, introduced this past April, and Fable 5 tell CyberScoop those models have been largely helpful to their work, even as they complained about high token usage and safety guardrails that hinder,  but don’t meaningfully prevent, defensive cyber tasks.

Eyal Webber Zvik, chief strategy officer at Cato Networks, a cloud and cybersecurity network provider in OpenAI’s Trusted Access in Cyber program, said they use GPT 5.5 and later OpenAI models to scan and triage internal codebases for vulnerabilities, test new safeguards and provide “highly autonomized service” to their customers.

Zvik wouldn’t disclose how many bugs 5.5 has found but said the company’s view is that it helps both find bugs that humans missed and rank which ones to patch based on factors like each bug’s exploitability.

“It is now a native part of our development environment and cycles, and we use those models to scale our entire codebase and make sure what we release into the service that our customers use to run their networks and network security has the least likelihood of having any vulnerabilities that can be exploited,” said Zvik.

John Hopper, vice president of engineering at SpecterOps, an identity security company, said newer models like GPT 5.5 are sharper and more persistent in pursuing their tasks.

“That can be a good or bad thing,” he noted.

One metric that SpecterOps tracks is how long it can keep a particular agent working before it moves off task or fails. That metric “matters a lot” because the longer an agent works without human help , the more agents a single operator can run at once.

Hopper said this provides defenders with immense value, and pushed back on the idea that the offensive capabilities the models offer are automatically more beneficial to malicious hackers. There is “a modicum of grounding that the industry needs when we talk about these models.”

“Yes, AI frontier tools will lower the barrier of entry, but these problems have always existed,” he said. “I don’t actually believe that AI is going to remove the needle in the haystack problem, but by howdy, using my two hands to find that damn needle, compared to using a backhoe, I can tell you which one I’d rather be driving.”

Eran Kinsbruner, vice president of product marketing at software security firm Checkmarx, told CyberScoop that later models like OpenAI’s Codex Security and GPT 5.5 are noticeably easier to set up and run with local systems, even for less technical users. That alone gives them an edge over many cybersecurity tools where interoperability is a constant concern.

However, GPT 5.5 burns through tokens at a much faster rate. He recalled one instance of using it to scan a medium-sized repository in three different programming languages.

“After 26 minutes I almost ran out of tokens, and it didn’t provide anything, just created a threat model for me and told me you want to buy more tokens?” he said.

In other instances, some of the scan results he received were not comprehensive.

Further, he expressed frustration with some of the guardrails designed to prevent risk – like only allowing users to scan local files but not code repositories like GitHub – “makes not too much sense” given how often developers must work with remote code.

Those kinds of guardrails – which can prevent models or developers from injecting malicious code or prompting into their models – sit at the heart of the debate in Washington D.C. and around the world. Some users feel differently about their utility.

Kinsbruner said that doesn’t make sense for organizations like his, which work with thousands of different enterprise organizations with  thousands of different code repositories spread across the internet.

“I cannot imagine how large-scale developers could just jump into this solution and make it an enterprise-grade, enterprise-level, de facto cybersecurity solution” out of it, said Kinsbruner.

OpenAI did not respond to a request from CyberScoop for an interview on GPT 5.5. The company has since released another model, GPT 5.6, that they said is more efficient at token use.

The White House’s crash course in AI cyber risk 

 The White House keeps changing its line on whether and how the U.S. government should limit the release of commercial frontier models. The shift comes from lessons learned since coming into office in Jan. 2025. Trump threw out Biden-era regulations meant to steer the industry toward safer models. Top officials like Vice President JD Vance argued against restricting industry progress.

Less than two years later, administration officials worry about the impact of speed and scale – two things AI excels at – in cyberspace.

According to Will Loucks, senior director of intelligence at the Office of the National Cyber Director, over the past two years the number of exposed and known vulnerabilities has shot up. Threat actors exploit those flaws faster before defenders can fix them. Once inside, the time from initial access to full network control shrinks.

“So in other words, every stage of the cyber operations lifecycle that a threat actor has to move through to get to a victim network and achieve an outcome, they’re just moving through more quickly faster,” said Loucks at a July 16 event in Washington D.C.

Speaking about AI in particular, Loucks said one of the defining characteristics of the technology is its ability to lower barriers for threat actors.

“Sometimes speed and volume have a threatening aspect alone, even if sophistication isn’t quite increasing in the same way, and the reason for that is because it places pressure on defenders…to triage alerts more quickly,” he said.

Jordan Rae Kelly, former director for cyber and incident response on the White House’s National Security Council during Trump’s first term, told CyberScoop that the changes over the past two years reflect the lessons the White House has learned on the issue since returning to office.

In the early days of this administration, Kelly said, “there is a sense and a spirit that the Biden administration was limiting AI and there was a kind of a rip-it-all-off [attitude], everybody go and do whatever, we will be the biggest and boldest and brightest.”

“I love that talking point, but I think what you’ve seen is probably an education over the last 19 months, where people [in the White House] have said that’s a challenging premise to put into place, knowing about the potential downsides and capabilities,” she added.

Michael Daniel, former White House cyber coordinator under President Barack Obama, thinks the horse may already be out of the barn.

Daniel, now head of the Cyber Threat Alliance, a membership nonprofit group focused on cyber threat information sharing between industry and government, said his members report that AI is being used to do things “faster and at a slightly bigger scale” but aren’t yet seeing the flood of exploitation that analysts have warned about. Not yet.

“I think what we’re seeing right now [and] talking about is ‘okay, where are the step changes [in the cyber threat landscape] actually going to occur?” said Daniel. “Are we and when will we see the explosion in vulnerability reporting from these Mythos-like capabilities? That’s what’s really got their attention right now.”

But Mythos and OpenAI’s Daybreak models are restricted to select organizations, and neither has publicly released its most powerful cybersecurity models to the public. That dynamic won’t last.

The UK’s AI Security Institute estimates that open source and foreign LLM models are between 4-7 months behind frontier U.S. models. In that setting, it’s hard to stop the development of AI models worldwide through export controls or other limits.

“It’s not like we’re buying ourselves five to ten years on this,” he said. “We’re not, and so I’m not sure the impact on the defenders who are trying to obey the law is worth whatever small hiccup we cause for our adversaries.”

Kelly said there’s merit to the administration’s current position, even if it took time to get there. Many federal cybersecurity procedures that operated even a decade ago – such as a Vulnerabilities Equities Process that could take days or weeks to consider the pros and cons of keeping an exploit – are no longer practical.

“All of that work to some degree, is out the window, because you can’t meet with the regularity you would need to meet to adjudicate vulnerabilities that are being found in seconds and exploited in minutes,” said Kelly.

But Kelly and others say that’s also because AI capabilities in cybersecurity are developing faster than policymakers can react, even in the best of times.

Key questions remain and the administration’s balance between national security and backing domestic industry will likely shift  in response to new events.  The administration wants a framework that can predict and manage the risks of AI models today and tomorrow. That may be harder than it sounds.

“Do I think they’ve been clear? No,” said Kelly. “But I think it’s a place where clarity is really hard to achieve.”

The post Where’s the Trump administration line on AI regulation? appeared first on CyberScoop.

Anthropic has announced its Fable 5 and Mythos 5 models will once again be available to the public as it has reached an agreement with the Commerce Department to deploy the AI models with new guardrails and classifiers meant to address jailbreaks.

In a blog posted Tuesday, Anthropic said that export controls that prevented their sale to foreign companies and individuals have been lifted after weeks of negotiation with the White House and Commerce Department. The company has also restored access to the model for U.S. users.

The export controls were put in place after the Trump administration became alarmed by a threat intelligence report from Amazon claiming to have jailbroken Fable’s cybersecurity capabilities.

On X, Secretary of Commerce Howard Lutnick appeared to confirm that the restrictions would be lifted.

“Over the past two weeks, we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US Government and strengthen America’s leadership in AI,” Lutnick wrote.

The administration levied the export controls after becoming concerned that the release of Fable 5 would lead to the model being jailbroken, giving users access to cybersecurity and other capabilities that Anthropic has said could wreak havoc on the open internet if  placed in the wrong hands. The Amazon report convinced administration officials that such jailbreaks were on the immediate horizon.

However, one oddity of the administration’s decision is that the capabilities described in the Amazon report, by all accounts, are not cutting-edge. Scanning code and breaking down how to exploit vulnerabilities for a user is already possible with existing models.

Anthropic confirmed that, saying that further testing found that equivalent and lesser models like ChatGPT 5.5, Claude Opus 4.8 and Kimi K2.7 could identify the same vulnerabilities as Fable did in the Amazon report, while a half dozen existing models were able to produce the same proof of concept code as Fable.

Crucially, Anthropic reiterated that they have yet to see a jailbreak that affects the model’s restrictions on cybersecurity and biology work, though they did call this instance “a borderline case.” Indeed, some cybersecurity professionals have publicly complained that existing safety guardrails on Fable 5 blocked many routine defensive cybersecurity work in addition to malicious use cases.

“Importantly, the reported technique did not expose any unique Mythos-level cyber capabilities,” the blog continued. “The behavior reflected a borderline case for Fable 5’s safeguards…there are some tasks that are unlikely to be dangerous but are nonetheless blocked by the safeguards out of an abundance of caution. The reported technique allowed access to one such behavior, but it only involved routine defensive cybersecurity work.”

Anthropic said it has trained new safety classifiers to target and block the behaviors described in the Amazon report and notify users when it happens, and that the new safeguards have been stress tested by the federal Center for AI Standards and Innovation. The new classifiers will block the techniques “99.9%” of the time, but Anthropic said they’re not expected to block all lower risk routine cyberdefense capabilities, just the most harmful ones.

The restrictions will likely make it even harder to use Fable 5 for defensive cybersecurity. One effect the company expects is that more “benign” requests for routine coding and debugging tasks will be flagged by the system.

Christopher Padilla, former Assistant Secretary for Commerce for export administration in the George W. Bush administration, said that while it’s “good news” the controls have ultimately been lifted, the Trump administration’s AI policy stumbles over the past two years illustrate “the risks of ad hoc, transactional policymaking.”

In a LinkedIn post, Padilla called the Trump administration’s approach chaotic and unpredictable — the opposite of the clear, consistent rules industry depends on. While Vice President J.D. Vance mocked AI safety regulations in a speech in Europe last year, the administration has quietly partnered with OpenAI and Anthropic on voluntary national security testing, especially as frontier models began showing advanced automation and cyberattack capabilities.

That national security arrangement was supposedly codified in a White House executive order last month, shaped heavily by industry boosters who feared regulatory delays would slow U.S. development. But days after Fable’s release, Commerce imposed new export controls on Anthropic’s models anyway.

Padilla called proposed AI safety regulations by the Biden administration “flawed and overly complex” but nevertheless predictable compared to the status quo. Instead of replacing those proposed regulations with their own vision, the Trump White House has been “to put it mildly, all over the place on AI policy.”

“The same BIS that stopped Fable and Mythos has a permissive policy for exporting high-end AI semiconductors to China — in exchange for a cut of the take,” said Padilla, referencing the Trump administration’s lifting of export controls on advanced AI chips. “This is not a smart way to make policy. Bad for industry competitiveness and for national security.”

The post US lifting export control restrictions on Anthropic’s Mythos, Fable appeared first on CyberScoop.

Intelligence agencies for the United States, Canada, UK, Australia and New Zealand are warning that advanced AI models capable of wreaking havoc in the cyber domain are “months away” from being publicly available.

In a joint statement, the Five Eyes alliance say they expect the kind of advanced hacking capabilities provided by frontier models like Anthropic’s Fable 5 and OpenAI’s Daybreak to become broadly available the public within the year, despite efforts by AI companies to withhold them or restrict their access.

“Frontier Al models are anticipated to exceed current industry expectations, fundamentally transforming both offensive and defensive cyber capabilities,” the agencies said. “The timeline is not years, it is months.”

The statement, which included signatures from NSA’s Director of the Cybersecurity Directorate David Imbordino and acting CISA Director Nick Andersen, does not specifically cite secret or classified sources or methods to reach this conclusion.

But much of the underlying justification provided by the intelligence agencies also aligns with what public cybersecurity and AI experts have been warning about for months.

AI models capable of exploiting cybersecurity weaknesses are already available today through multiple channels: older commercial models, open-source versions, or foreign and black-market sources. And while newer models like Mythos are reportedly significantly more powerful for cybersecurity-related tasks, the breakneck pace of frontier model development often means that yesterday’s restricted frontier AI is tomorrow’s free, open-source AI.

Representative Andrew Garbarino, R-N.Y., Chair of the House Homeland Security Committee, said the warning from intelligence agencies “underscores what the Committee has repeatedly heard through roundtables, briefings, and hearings with industry leaders: China is just months, if not now weeks, away from achieving frontier AI capabilities comparable to those of the United States.”

“This threat reinforces the urgency of ensuring that federal agencies and critical infrastructure operators can responsibly leverage advanced U.S. models, and receive the guidance and support necessary to do so, to find vulnerabilities before adversaries can exploit them,” said Garbarino in a statement.”

The agencies flag legacy systems, sluggish patching loops, unnecessary internet connectivity, weak identity and access controls, and a lack of pre-incident planning by organizations as key weaknesses that AI will excel at exploiting.

“The rapid pace of frontier AI development means cyber risk assumptions can become outdated in months, not years,” the agencies wrote. “We must act before and be prepared to adapt and withstand evolving threats.”

Since large language models burst onto the scene, open-source models have run about 6-8 months behind the largest frontier AI companies.

To give an idea of how quickly the field develops: the capabilities described in the Amazon threat intelligence report that convinced the Trump administration to place export controls on Fable 5 could already be accomplished through older models like Claude Opus and Claude Sonnet, as well as open-source Chinese models.

Anthropic shut down access to their Fable 5 and Mythos 5 models as a result, and despite releasing a statement that they believe the White House decision was a “misunderstanding” the dispute remains resolved.

Programs like Anthropic’s Project Glasswing and OpenAI’s Trusted Access for Cyber Program provide AI systems to organizations for cyberdefense.  The goal is to give defenders a head start in finding and fixing vulnerabilities before AI systems can exploit them routinely in the coming years.

However, for all the fear surrounding the new technology, the recommended guidance is largely the same as it has been for decades. Governments, businesses and leaders must stop treating the digital security of their work as an afterthought or compliance issue.

“Success will come from getting the basics right, acting quickly, and integrating cyber security into core business strategy,” the agencies wrote. “Those that do not will face growing operational and strategic disadvantage.”

06/23/2026: This story was updated to include comment from Rep. Andrew Garbarino, R-N.Y.

The post Intel agencies: Frontier AI models will reshape cybersecurity faster than expected appeared first on CyberScoop.

Members of Congress responded with skepticism and caution Tuesday to the Trump administration’s decision to impose export controls on Anthropic’s newest AI models.

The Friday order, which Anthropic said forced it to disable its Fable 5 and Mythos 5 artificial intelligence models, was prompted by what the administration said were national security concerns that a large number of cybersecurity professionals have dismissed as ill-founded.

Several Hill Democrats told CyberScoop they were concerned that the administration’s decision was driven by other considerations. Notably, the administration has feuded with Anthropic over use of its models for domestic surveillance and fully autonomous weapons.

Sen. Angus King, a Maine independent who caucuses with Democrats, said he would need to be convinced it was a legitimate national security order and hadn’t yet seen a full justification.

“What they did was pretty extreme, and I’d want to see what the basis was, as opposed to all the other issues that are swirling around in cybersecurity,” he said. “I’m a little skeptical because of their otherwise announced antipathy to this company.”

Leaders of the House Homeland Security Committee had contrasting takes, with Chairman Andrew Garbarino, R-N.Y., offering a two-pronged response and the top Democrat on the panel, Bennie Thompson of Mississippi, panning the order.

“The administration is right to treat advanced AI cyber capabilities as a national security issue, especially when foreign adversaries and cybercriminals are actively looking for ways to weaponize these tools,” Garbarino said in a statement. “At the same time, we need to make sure our response does not unintentionally disadvantage American companies, allied partners, or critical infrastructure defenders who need access to the best secure tools available in order to protect our networks here at home.”

The United States, not China, needs to set standards for trusted AI, Garbarino said.

But Thompson said the order adds evidence to the appearance that the Trump administration doesn’t “have a coherent plan for mitigating the cybersecurity risks” of frontier AI models, he told CyberScoop in a statement.

“AI regulations should rely on standards and procedures that provide confidence to the public that decisions are based on the evidence and not on politics,” he said. “Instead, the Trump administration has adopted an ad hoc approach where decisions are made by political appointees in the White House rather than experts and where companies are left guessing on how to comply.”

Virginia Sen. Mark Warner, the top Democrat on the Intelligence Committee, had also previously highlighted the administration’s quarrel with Anthropic in response to the order in a statement to CyberScoop.

Behind the scenes, the administration and Anthropic were reportedly continuing to try to forge a truce Tuesday. More broadly, the administration’s AI executive order had a rocky rollout as the administration swung back-and-forth on how involved the government should be.

Some lawmakers deferred on commenting Tuesday, such as Senate Homeland Security Committee Chairman Rand Paul, R-Ky., who told CyberScoop he didn’t have anything to say on the order.

Others said they were still seeking information from the administration.

“I have not had the opportunity to get a brief specifically as to the logic, the reasoning behind it, and so forth,” said Sen. Mike Rounds, the South Dakota Republican who chairs the Armed Services Subcommittee on Cybersecurity. “So I’m going to withhold judgment until I get an opportunity to get the rest of the story, so to speak.”

The post Lawmakers leery about Trump administration’s Anthropic order appeared first on CyberScoop.

Last Friday, the Trump administration sent a shock through the tech ecosystem when the Department of Commerce levied export controls on Anthropic’s new AI model Fable 5.

Anthropic has taken steps to limit the risks around the commercial sale of its Mythos model, including declining to release it publicly, funneling it to organizations for cyber defense and developing guardrails for Fable 5 that would default its answers to older, less powerful models around sensitive topics like cybersecurity and biological warfare.

But the Trump administration was reportedly alarmed by recent reports from Amazon and another cybersecurity researcher claiming to have jailbroken Fable 5 within days of its public release, and determined that if researchers in the U.S. could jailbreak the model, so could America’s foreign adversaries.

The Commerce Department’s decision spurred Anthropic to shut off the models for all users as they attempted to convince the White House to change course.

But some cybersecurity and AI experts have sharply disagreed with the White House’s actions, saying the research has not demonstrated that anyone has been able to circumvent Fable 5’s safeguards and access the kind of dangerous new capabilities that have worried officials.

Katie Moussouris, a well-known cybersecurity expert, said Monday that Anthropic provided her with a copy of third-party research on guardrail bypass techniques for Fable 5.

According to Moussouris, the researchers asked three Claude models – Fable 5, Mythos and Claude Opus – to review batches of known, vulnerable open source code for security issues. Fable 5 initially refused the request, but the researchers were able to use “a multistep and manual process” to get Fable 5 to turn the output into automated scripts that could test patches for the vulnerability.

Third-party research since Fable 5’s release has not found ways to bypass its safeguards around hacking. The capabilities researchers have demonstrated are foundational to what makes Fable 5 and other frontier models valuable for cybersecurity defense.

“Defenders need to be able to ask AI to fix the bugs in a file, explain why the fix matters, and write tests that confirm the patch works,” she wrote. “That is not a guardrail bypass. It is the most valuable thing an AI model can do for defensive security: executing the find, fix, and test loop defenders run every day.”

Moussouris previously provided technical expertise to the Waasenaar Agreement, a voluntary multilateral security agreement around controlling exports for both munitions and dual use technology that includes the U.S. and dozens of other countries.  Based on the research she’s seen, she called placing export restrictions on all foreign sales of Fable 5 “heavy handed” and “misguided.”

Some lawmakers who in favor of higher regulations and scrutiny on the national security implications of AI were nevertheless critical of the White House decision. Senator Mark Warner, D-Va., told CyberScoop in a statement that while “there may be circumstances where restrictions on the export of frontier AI models are warranted,” those decisions must be “grounded in a transparent, risk-based process with clear rules and consistent standards.”

The Trump administration’s approach, he argued, has been the opposite, and he called for Congress to pass a statutory framework for testing and approving frontier AI models based on transparency, predictability and fairness.

“This administration has repeatedly shown a willingness to weaken export controls designed to protect our national security and maintain our technological edge over adversaries, while also making no secret of its hostility toward Anthropic,” said Warner. “That raises serious questions about whether this effort is being driven by objective national security concerns or something else.”

Anthropic said it subjected Fable 5 to 1,000 hours of testing from internal and external red team, reporting that no universal jailbreaks were found that would remove those guardrails or allow the model to access Mythos for cyber and biology work.

Moussouris is far from alone. She is one of dozens of cybersecurity experts who signed an open letter Monday calling on the Trump administration to “Free Fable.”    

The researchers say that while Mythos-class models are “quite good” at identifying and exploiting vulnerabilities in software code, they “are not uniquely good” compared to other frontier models they use every day for cybersecurity defense.

For example, despite OpenAI’s Daybreak model offering similar vulnerability discovery and patching capabilities. It was not included in the Commerce Department’s restrictions.

The researchers also note that Fable 5’s guardrails have been notoriously oversensitive compared to other frontier models used by red teamers, becoming “a source of humor in the cyber community on launch day” as IT and cyber workers reported online that they couldn’t get the model to perform basic defensive cybersecurity tasks.

The letter questions whether the issues found in the jailbreaking reports would even qualify as offensive capabilities, and note they can be reproduced in other commercial and open-source models, including GPT 5.5, Claude Opus, Claude Sonnet and Chinese models like Kimi 2.7.

“The justification for this unprecedented action was that Fable provides a unique ‘uplift’ of capabilities beyond other AI models, but AI has been finding bugs and generating working exploits at superhuman levels since last year,” they wrote.

The White House decision comes as AI companies face increasing backlash from a public that is now overwhelming calling for more robust government intervention.

A Johns Hopkins University poll in May found broad, bipartisan support for AI regulations, with 73% calling for bans on AI-generated images and video, 68% calling for labels on AI content, 75% wanting disclosure laws around when they interact with AI chatbots and 70% calling for “the right to interact with a human rather than an AI in medical, legal, educational and government settings.”

Another global survey of 18,000 people released this week found that the top four concerns most people have around AI all revolve around the tool’s ability to spread misinformation, create deepfakes to embarrass or hurt others, making it easier for criminals to hack into victim networks and helping terrorists create new weapons.

Senior reporter Tim Starks contributed reporting for this story.

The post Cybersecurity experts don’t think Anthropic’s Fable 5 presents a unique threat  appeared first on CyberScoop.

The U.S. government on Friday ordered Anthropic to immediately suspend foreign access to Fable 5 and Mythos 5, its two most advanced artificial intelligence models, citing national security concerns tied to a reported method of bypassing the models’ safety restrictions. 

The directive, issued late Friday afternoon by Secretary of Commerce Howard Lutnick in a letter to Anthropic Chief Executive Dario Amodei, placed the two models under export controls that prohibit use by foreign nationals, whether inside or outside the United States. 

Because of the scope of the restrictions, which includes foreign-born Anthropic employees, the company announced Friday evening that it disabled the models to ensure compliance. Access to the company’s other AI models was not affected. 

Fable 5 and Mythos 5 had been released earlier this week, with Anthropic describing them as the most capable systems it had ever deployed. Mythos was available to members of Project Glasswing, which allowed selected cybersecurity companies to use the model to identify and address security flaws.

It’s unclear how the Commerce Department action affects Project Glasswing. Anthropic did not respond to a request for comment.

The Commerce Department‘s letter did not detail the specific national security concern. In its blog post Friday night, the company said its understanding is that the government became aware of a technique for “jailbreaking” Fable 5, a term for methods that circumvent a model’s built-in safety guardrails. According to Anthropic, the government provided only verbal evidence of what it described as a “narrow, non-universal jailbreak,” which essentially involved prompting the model to read a specific codebase and identify software flaws. 

Anthropic disputed the severity of the finding. The company said it reviewed a report it believes formed the basis of the government’s directive and found that the capabilities demonstrated were already available in other publicly accessible models, including OpenAI’s GPT-5.5. The company said those same capabilities are used routinely by cybersecurity professionals for defensive purposes. 

Katie Moussouris, chief executive of the cybersecurity firm Luta Security, posted on BlueSky Saturday that the issue stems from “Defense Oriented Prompting,” a security-first method of engineering AI system instructions that treats natural language as code.

Other reports claimed that Amazon was responsible for flagging the security issues in the model. The company did not respond to CyberScoop’s request for comment. 

Anthropic acknowledged in its statement that perfect jailbreak resistance is not achievable for any model provider, and said it had designed Fable 5 around a “defense in depth” strategy, combining narrow jailbreak resistance with active monitoring. The company said no testers had found a universal jailbreak capable of broadly bypassing the model’s safeguards. 

“We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,” Anthropic wrote. “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”

Friday’s directive is the latest episode in a prolonged dispute between Anthropic and the Trump administration. In February, President Donald Trump moved to bar Anthropic’s products from federal agencies after the company sought stronger restrictions on how the Pentagon used its technology.

Despite that, as Anthropic released Mythos under Project Glasswing, the National Security Agency was given Mythos 5 to conduct offensive cyber operations. Earlier this month, Trump signed an executive order directing federal agencies to bolster cyber defenses and establish a voluntary mechanism for the government to gain early access to powerful AI models before deployment. 

The administration’s stated rationale for Friday’s action drew widespread skepticism from researchers and analysts. Dean Ball, a senior fellow at the Foundation for American Innovation, called the move “baffling.” Chris McGuire, a senior fellow at the Council on Foreign Relations, said targeted export controls on model access could be a legitimate policy tool, but called the across-the-board restriction “highly questionable” and the deemed export provisions — which restrict foreign nationals inside the U.S. — “just absurd.” 

The broader implications for the AI industry remain uncertain. Aaron Levie, chief executive of Box, described the directive as “a big turning point for AI regulation,” arguing that the government’s willingness to deem specific models too powerful for certain uses establishes a precedent with potentially far-reaching consequences.

Other tech leaders in the government supported the action. 

“We fully support @POTUS and @SecWar in prioritizing national security and the security of our warfighters, DIB partners, critical infrastructure, international partners and allies,” DOD CIO Kirsten Davies wrote in a social post on X. “Some things are simply more important than revenue cycles, clickbait, and pre-IPO valuation. America First. Always.”

Anthropic said it believes the situation stems from a misunderstanding and is working to restore access as soon as possible.

The post Anthropic disables new models after government calls them a national security concern appeared first on CyberScoop.

Earlier this year, Anthropic executives said that their new AI model, Claude Mythos, had such powerful capabilities for harm that they would not release it publicly.

On Tuesday, the company said it was making an altered version of Mythos available to the public, promising “new guardrails” that thwart the model’s best-in-class performance in hacking and bioweapons research.

Anthropic said Claude Fable 5 was the “same underlying model” as Mythos, but its responses for certain topics like cybersecurity and biology will be drawn from a previous Claude Opus model that is already public.

“Releasing a model this capable comes with risks. Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage,” the company said in a draft blog sent to CyberScoop ahead of the announcement. “We’ve therefore launched the model with safeguards that route queries on a narrow set of topics to our next-most-capable model, Claude Opus 4.8.”

Anthropic also said they subjected Fable 5 to both internal and external red team testing for common model vulnerabilities, like jailbreaking. Anthropic said these tests identified no known “universal” jailbreaking techniques, but does not specify if partial jailbreaking techniques were discovered.  

The company is betting that won’t change when Fable 5 is made available to the broader public, but it’s worth noting that cybersecurity researchers have consistently found ways to jailbreak older AI models.

“The uplift from Mythos-level capabilities is valuable to many adversaries—for instance, those who could financially gain from cyberattacks—and we therefore expect them to be motivated to try to circumvent our safety measures,” the company wrote.

Anthropic is changing its data retention policies for Fable and Mythos models, keeping all user traffic for 30 days on both its own platforms and third-party services. A White House executive order creates a voluntary framework for AI companies to share frontier models with the government up to 30 days before public release. The company says the retained data won’t be used to train new Claude models or for “any non-safety-related-purpose.”

Following publication, a spokesperson for Anthropic told CyberScoop the company’s data retention policies “are specific to their safeguards work and is unrelated to the EO.”

Most organizations are still deciding whether to adopt AI into their IT and cybersecurity ecosystem.  But models like Mythos can scan for vulnerabilities, chain together exploits, and steal data from a victim network in minutes. Automation in hacking existed before AI, but experts have said frontier models like Mythos and OpenAI’s Daybreak can allow even low-level cybercriminals to wreak havoc.

While Anthropic cited its commitment to developing safe and secure AI in its reasons for not publicly releasing Mythos, many organizations have been clamoring for access, and its enhanced cybersecurity functions in cybersecurity and other areas have been the subject of congressional hearings, national security papers and White House executive orders.

Releasing a limited version of the model in Fable 5 represents an attempt to split the difference between those two desires. Anthropic said it would release follow up benchmarks and assets for the model.

So what can Fable 5 do? 

Anthropic said it’s possible the restrictions built into Fable will make it harder for the model to fulfill both malicious and legitimate user requests.

“Because we have prioritized safety, we’ve deliberately tuned the safeguards to be cautious, and they are still stricter than would be ideal—for example, sometimes benign requests will trigger our classifiers,” the company wrote. “We recognize that this will be frustrating to some users, and our aim is to reduce false positives as we update and refine the safeguards after launch.”

If Fable 5 draws its cybersecurity and biology answers entirely from Claude Opus 4.8, it will still provide users with impressive – though not unique – dual use cybersecurity capabilities.

According to the system card published for Opus 4.8, the model is a slight improvement on previous models like 4.7 in the realm of cybersecurity but was “generally much less capable than Mythos Preview.”

Opus 4.8 was tested on its ability to write complete end-to-end exploits and build exploit primitives that provide attackers with the ability to execute arbitrary code. It averaged a score just 5 out of 16 in proficiency, compared to Mythos Preview which scored closer to 10.

Without safety guardrails in place, Opus 4.8 can still reproduce nearly 80% of previously discovered vulnerabilities in real open-source software projects when given a high level description of the weakness. The system card said Anthropic’s unspecified safeguards whittle this success rate down to 1%.

Another test assessing Opus’ ability to develop exploits for the popular Firefox browser found that, again without guardrails, the model could identify a full working exploit 8.8% of the time and a partial working exploit 68.8% of the time.

The company also said that members of Project Glasswing – a consortium of public and private businesses given access to a preview version of Mythos – will be able to upgrade to the latest full model, Claude Mythos 5, to continue their work. Access to Mythos 5 will be expanded over time “through a more systematic trusted-access program” including federal agencies.

The post Anthropic’s new model is Mythos on a leash appeared first on CyberScoop.

AI models such as Anthropic’s Claude Mythos and OpenAI’s Daybreak represent a fundamental inflection point in security. These advances are not only reshaping technology but also redefining trust, risk, and the relationship between humans and intelligent systems. As innovation accelerates, AI governance and responsible deployment are becoming strategic priorities for every organization.

Historically, governments have played a stabilizing role during moments of transformational technological change. Yet the pace and scale of the AI era demand a new model, one built on partnership rather than control, balancing societal responsibility with the need to sustain innovation and global competitiveness.

The White House’s executive order on AI governance signals that collaboration between the industry and policymakers will increasingly shape the future landscape. Proposed frameworks that promote transparency and responsible development point toward a more coordinated approach to risk management.

Effective governance of AI models should balance clear safeguards with the speed of innovation, aligning organizations, policy makers, and technology leaders around a shared goal: advancing AI in ways that strengthen trust, security, and long-term value. The path forward is not defined by heavy-handed oversight, but by building an ecosystem of accountability.

Three key points substantiate this approach.

First, the industry should recognize Anthropic’s release of Mythos as an example of responsible innovation. Company leaders recognized the model’s risks and deliberately delayed broader deployment, allowing early testing to surface vulnerabilities before widespread adoption.

The broader lesson extends beyond a single model release. Responsible leadership means prioritizing decisions that build trust and enable sustained innovation. As AI capabilities accelerate, the most successful organizations that lead will be those that weave accountability through their ambitious pursuits, rather than treating them as competing priorities.

Second, innovation rarely thrives under rigid frameworks. History has shown that many compliance regimes, while well-intentioned, incentivize organizations to optimize for requirements rather than outcomes. Security is strengthened through systems designed for resilience and trust, which goes beyond mere compliance.

Third, slowing U.S.-based AI innovation risks weakening long-term competitiveness. The U.S. remains a leader in AI but maintaining that position will require balancing responsible safeguards with continued investment and progress. Overly restrictive approaches risk slowing domestic advancement while other nations continue accelerating development and capability.

An effective AI governance approach would encourage further responsible AI model development, as demonstrated by Anthropic. It would avoid direct government regulation and instead enforce accountability for companies that are irresponsible with AI development.

Hopefully, the partnership and collaboration between government entities and industry will continue beyond the White House order. Policymakers and industry leaders should create incentives that reward AI vendors for considering societal implications before releasing new solutions. This framework would highlight responsible providers as models for the industry while imposing meaningful consequences based on demonstrated societal harm that direct affects business and technology decisions.  

AI models such as Mythos and Daybreak underscore a broader reality: the future of AI will be shaped by the trust around innovation, not merely by its development pace. The next era of AI leadership will require a new model of collaboration between industry and policymakers that maintains the speed and adaptability that innovation demands while establishing meaningful accountability for real-world outcomes.

The objective should be to guide progress responsibly. The organizations and nations that lead in the AI era will be those that demonstrate how innovation and accountability work together to strengthen trust, security, and long-term value creation.

Art Gilliland is CEO of Delinea, a cybersecurity company focused on human, machine and AI identity protection.

The post The AI security race needs accountability, not overregulation appeared first on CyberScoop.

Troy West was in Warsaw when his dinner was interrupted by his phone. But he was happy about it.

West, associate director of cybersecurity for autonomous offensive security company XBOW, had just learned that a trial version of the company’s platform had found a vulnerability that led to a full takedown of a development environment used by Moderna, the pharmaceutical company primarily known for its work related to mRNA vaccines.

It was, by most measures, exactly the kind of outcome a security team dreads. But for West and Farzan Karimi, Moderna’s deputy CISO, it was something closer to a proof of concept. XBOW’s product had done in hours what a human penetration tester could not — and it had done so with a level of persistence and creativity that neither of them had fully anticipated.

The episode is one data point in a much larger shift now rippling through the cybersecurity industry: The artificial intelligence models discovering vulnerabilities are moving faster than the teams that have to patch them.

Across recent conversations and presentations, industry experts said the tools are getting sharper, the attack surface is getting larger, and the gap between finding a problem and fixing it is not closing fast enough. For now, most organizations are caught between the speed of discovery and the slowness of remediation, with vendors across the industry rushing to position their products as the way through.

A shift in scale 

The inflection point came with Claude Mythos. When Anthropic announced the highly guarded model, security executives at major enterprise technology companies took notice in a way they had not with prior frontier releases. 

Zscaler was among the early organizations given access to the model, and CEO Jay Chaudhry told CyberScoop that he directed his team to use it to probe the company’s own applications.

“Are we finding some serious stuff? Yes, indeed,” Chaudhry told CyberScoop at Gartner’s Security & Risk Management Summit. He was careful to note that the findings were not necessarily more severe than those produced by other models. The issue, he said, was volume. 

“There aren’t enough resources and cycles to fix all those,” he said. 

The reason Mythos changed the calculus, according to Tom Gillis, general manager for infrastructure and security products at Cisco, comes down to code complexity. Legacy network infrastructure was built on tens of millions of lines of code developed over decades, and earlier AI models lacked the context window and reasoning capacity to comprehend it in full.

“The models couldn’t understand the entirety of it before,” he told CyberScoop. “Now they can. That’s why they’re finding all these vulnerabilities.”

The problem runs deeper than application code. Firewalls and network switches often run for decades without updates or reboots, and many have never been patched in any meaningful way. The combination of aging infrastructure and newly capable AI models has created what Gillis described as a meaningful and accelerating shift in attacker capability that the industry’s existing operational rhythms were not built to absorb.

An opportunity in existing technology 

Cisco’s answer to the oncoming vulnerability deluge is a technology it calls Live Protect, a compensated control built on eBPF, a Linux feature that lets security software operate at the kernel level to block threats without rewriting system code.

“It’s a pinpoint, laser-fine control that can shield a vulnerability on a production system,” Gillis said. “We’re not touching or modifying the binaries of that production system.”

The intent is to shrink the window between discovering a vulnerability and the next scheduled patch, allowing IT teams to fix issues without taking systems offline.

“This is a finger in the dike that plugs a hole until you get to new change control windows,” he said, acknowledging that some customers may be tempted to treat the shields as a permanent solution. 

The product has been shipping since October, but customer urgency shifted noticeably after Mythos. “Customers are like, ‘Oh, good story, Tom. I’ll think about it.’ Now it’s like, ‘Oh my God, turn this thing on right now.’”

He also noted that eBPF is open source, and said he expects the broader industry to follow. 

“While I’m very proud of Cisco leading the market with these compensated controls, I know my competitors have to do this.”

The bot that broke everything 

But shielding vulnerabilities only works if you know they exist. Karimi, the Moderna deputy CISO, faced a different problem: His vulnerability management system was surfacing hundreds of high-severity findings with no reliable way to know which ones an attacker could actually exploit. His team had skilled red-teamers, but they were finite resources. What he needed was something that could test continuously, everywhere.

“We have some very senior red-teamers and pen-testers in our organization that are pointed in a specific direction,” Karimi said during a presentation at the Gartner summit. “XBOW is covering different attack stories for us.”

West, who leads offensive security for XBOW, describes the platform as a response to a structural problem in how offensive security has traditionally worked. Human testers scope an engagement, run it, write a report, and move on. The window between tests is where risk accumulates.

“Historically you have exploit developers spending time finding the right vulnerabilities, writing the exploits, finding if those exploits are reachable, and then finding a way to chain them all together,” West said. “That takes a long time.”

Given the realities, Karimi decided to put XBOW through a trial, which produced two notable findings.

In the first, XBOW identified a web application firewall bypass on a company application built on the Spring Boot framework. The bypass involved encoding a single character (a capital “A”) as its percent-encoded URL equivalent (A), which the WAF interpreted as a legitimate request, allowing the bot unfettered access. 

The second finding, which was the cause for West’s dinner interruption, was more consequential. West had provided XBOW with access to the source code of an internal application called Orders, used by Moderna’s research partners to procure drug substances, but no login credentials. The platform identified a valid API key embedded in the source code, used it to authenticate, and then began probing the application’s APIs for SQL injection vulnerabilities.

What happened next was not entirely planned. One of those APIs handled a malformed SQL injection attempt in an unexpected way, dumping garbage data into a shared routing application that other services depended on.

“Not only was it able to kick that Orders app I showed you, but it somehow kicked over the entire ecosystem of apps,” West said.

Human pen-testers who reviewed the findings afterward confirmed they were valid, and said they would not have found them on their own. Karimi said despite the outage, his team recognized the value immediately.

“If we’re able to demonstrate where you could have an outage in a safe testing environment, that’s a great signal,” he said.

The broader value, Karimi argued, is in forcing prioritization when bugs are discovered. “If you have exploit proofs, you can provide that plus-one modifier and really point your developers to remediate the top tier of real risk that’s been validated.”

But he does worry about the volume of bugs that will be surfaced by these tools. 

“How do we now handle the volume of bugs that have gone up due to AI-driven scale?” he said. “That’s a whole other problem space.”

A broader reckoning

Across these conversations, a consistent theme was that even as defenders are trying to get arms around the forthcoming wave of bugs, it’s going to be a tremendously uphill battle. That mirrors what some of the industry’s top leaders have been saying for months. 

It also mirrors what the model developers themselves have consistently been warning about. In its announcement about expanding access to Mythos, Anthropic admitted the timeline for a publicly available tool similar to its cybersecurity-focused model is shortening, and there are no guarantees it will be released with safeguards. 

“In that world, cyberattacks could occur much more often, and in much more unpredictable forms,” the blog post reads.

Gillis was blunter about what happens to organizations that don’t move. 

“Some people will be slow to change,” he said. “But the consequence of not making that change is gonna be front-page news. It’s a massive, massive compromise. You know, like, ‘you gave up every credit card number.’ Bummer.”

The post Inside the race to adapt to an AI-powered security world appeared first on CyberScoop.

Anthropic is broadening access to its Project Glasswing program, adding approximately 150 organizations in 15 countries, the company announced Tuesday, as its restricted Claude Mythos Preview model has already surfaced more than 10,000 high- or critical-severity software vulnerabilities since the program launched in early April.

The expansion follows an initial cohort of roughly 50 partners that were announced when Anthropic first unveiled the initiative. Those members included technology companies such as Amazon Web Services, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks, among others.  

According to the announcement, the new group covers sectors that were underrepresented in the first wave, including power, water, healthcare, communications, and hardware. Many of the new partners are vendors whose codebases underpin critical infrastructure systems.

The company did not give any further details on what companies or organizations were part of the new cohort.  Sources tell CyberScoop that NetSkope and Rubrik, which specialize in cloud security and data management, is part of the group given access in this latest round.

The scale of what Mythos Preview has already found is drawing attention across the security industry. Cloudflare identified 2,000 bugs across its critical-path systems, including 400 rated high or critical, with a false-positive rate the company described as better than that of human testers. Mozilla found and fixed 271 vulnerabilities in Firefox 150 while testing the model, more than 10 times the number found in a previous Firefox version using an earlier Anthropic model. Several other partners reported that their rates of bug discovery increased more than tenfold after deploying the model. 

Anthropic also used Mythos to scan more than 1,000 open-source projects, flagging 23,019 potential vulnerabilities, 6,202 of them estimated as high or critical. Of 1,752 high- or critical-rated findings independently reviewed, over 90% were confirmed as valid. 

The findings have shifted what Anthropic describes as the central issue in cybersecurity. Despite the enhanced ability to discover flaws, the company admits there are challenges with verifying, disclosing, and patching them before attackers can take advantage.

“The bottleneck in fixing bugs like these is the human capacity to triage, report, and design and deploy patches for them,” the company said in its blog post. 

That bottleneck has broader implications. A joint report from the Cloud Security Alliance, the SANS Institute, and OWASP concluded that organizations are “likely to be overwhelmed” in the near term by threat actors using AI to find and exploit vulnerabilities faster than defenders can patch them.

Anthropic has said it will not release Mythos-class models to the general public, citing the absence of safeguards sufficient to prevent serious misuse. In the interim, it has released Claude Security, a product using its publicly available Claude Opus 4.8 model that has been used to patch more than 2,100 vulnerabilities in three weeks. 

The program’s expansion comes as the Trump administration signed a scaled-back executive order on AI security. The order, which was signed hours after Anthropic’s announcement, sets up a voluntary framework requiring AI developers to submit advanced models to a government review up 30 days before public release.

The post Anthropic expanding access to Project Glasswing appeared first on CyberScoop.

As defenders get their hands on newer AI models with more powerful cybersecurity capabilities like Anthropic’s Mythos and OpenAI’s Daybreak, organizations are being told to prepare for a flood of new vulnerability reports.

But for bug bounty programs across the nation, that day may already be here, as yesterday’s frontier models and today’s open-source AI tools have dramatically increased the volume of bug reports flowing into companies around their own products or on larger bounty platforms online.

GitHub, one of the world’s largest online code repositories, said it is tightening its definition of a “complete” bug report after a significant increase in AI-assisted submissions over the past year.

Although the influx has had some benefits, many reports are submitted without proof of concept, are reliant on unrealistic attack scenarios or cover issues already listed as ineligible. As a result, the company is having difficulty separating signal from noise.

“This isn’t unique to GitHub,” wrote Jarom Brown, senior product security engineer at GitHub. “Programs across the industry are grappling with the same challenge, and some have shut down entirely.”

Brown said GitHub does not want to ban the use of AI generated reports entirely, calling it a “force multiplier” for security in the right context. But in a world where it’s never been easier to use AI to generate theoretical bugs, the company wants researchers to go the extra mile to confirm that their discoveries can actually be exploited in real-world conditions.

What we need is the same standard we’ve always expected: validation,” Brown wrote. “An AI-assisted finding that’s been verified, reproduced, and submitted with a working proof of concept is a great submission. An unvalidated output submitted as-is without reproduction or demonstrated impact is not.”

Grant Bourzikas, chief security officer at Cloudflare, said triaging bugs and proving they can be exploited  has always been one of the hardest parts of vulnerability research, and AI vulnerability scanners and code have “made it worse.”

For instance, code written in C and C++ programming languages are vulnerable to a range of exploits – like buffer overflows and out-of-bounds reading and writing – that don’t exist in memory safe languages like Rust. AI tools scanning software written in memory unsafe programming languages are far more likely to generate false positives.

But one of the biggest flaws continues to be that AI tools are also designed to give the user what they’re asking for, even when it’s not there. This leads to the generation of bug reports filled with speculation and qualifiers around exploitability that require human follow up.

“That’s a reasonable bias for an exploratory tool,” Bourzikas wrote. “It’s a ruinous one for a triage queue, where every speculative finding spends human attention and tokens to dismiss, and that cost compounds across thousands of findings.”

Cloudflare recently shared results from testing Mythos on 50 of its own code repositories, looking for exploits. Bourzikas called Mythos “a different kind of tool doing a different kind of work” from other frontier models, and that it made significant progress in reducing false positives.

For example, he pointed to two Mythos capabilities that stood out compared to other models: chaining exploits together and generating its own proof-of-concept code to confirm exploitability.

Older models could spot many of the same bugs, but they often couldn’t figure out how to exploit them effectively, or show that the issue could be exploited in real world conditions.

Others have argued that the gap in bug hunting capabilities between newer frontier AI models and older ones, or open source models available today is not as large as advertised. 

Swedish software developer Daniel Stenberg, lead developer for curl, an open source file transfer tool used around the world, recently wrote about his experience with Mythos Preview. Like others, he has also seen a higher volume of AI-fueled bug reports over the past year, but said the flood of low-quality reports has tapered off significantly since March as models have improved.

Curl is mature and polished by the standards of most software: Stenberg estimates each line of code has been rewritten or altered at least four times, and he said he has used both human and AI tools in the past to implement hundreds of bug fixes over Curl’s existence.

That makes it a unique testing ground for the enhanced capabilities of Mythos, which was reportedly so powerful at finding vulnerabilities that Anthropic opted not to release it to the general public.

After gaining access to Mythos, Stenberg received the results of a scan of 178,000 lines of curl code. Ultimately, the scan flagged five “confirmed” vulnerabilities. Further exploration by human researchers found that 4 of the bugs were false positives or had no security impact. The one remaining bug Mythos found? A low-severity flaw that will be fixed in a regular June update.

Even as he praised the impact of AI on cybersecurity generally, Stenberg concluded that for all the hype, Mythos is only “a bit better” than previously released models.

“My personal conclusion can however not end up with anything else than that the big hype around this model so far was primarily marketing,” he wrote. “I see no evidence that this setup finds issues to any particular higher or more advanced degree than the other tools have done before Mythos.”

The post AI might cut false positives, but it won’t stop the slop  appeared first on CyberScoop.

Advanced artificial intelligence models will “fundamentally change warfare as we know it,” a top cyber official at the Defense Department said Thursday, saying it represents “not evolutionary warfare, but revolutionary warfare.”

Paul Lyons, principal deputy assistant secretary for cyber policy, said the development of frontier AI models like Mythos amounted to a “watershed moment,” speaking at Rubrik’s  Federal Cyber Resilience Breakfast produced by FedScoop.

Such models will “change both offense and defensive posture within the Department of War to something that’s close to you for critical infrastructure,” he said. “This is the ability to hunt and speed across the domain and outside the fence line in critical dependencies with water, power, compute.”

The advent of the technology is forcing the department to address difficult questions, but it’s a great opportunity as well for the United States given that it’s being developed by American companies, Lyons said. It’s something his department is optimistic about, he said.

“To be blunt, we’re trying to figure out, what authorities do we need? How do you leverage that within both decisionmaking and employment?” he said. “We have the right people looking at the speed, scale and complexity of cyber and how it’s going to be affected through the advent of AI.”

The Pentagon labeled Mythos a “supply chain risk” after its creator, Anthropic, resisted commands from the department to use its Claude model in ways the firm opposed. The department has nonetheless been using Mythos to hunt for cyber vulnerabilities.

Lyons said that cyber warfare overall has become more mature, as recent conflicts have shown.

“We saw it in spades in Venezuela, where you can layer cyber to create conditions that are favorable to the warfighter, that lower risk to mission, lower risk to force that where paired with both no kinetic and kinetic effects, can increase lethality,” he said. “We see it in Iran today.”

President Donald Trump’s cyber strategy places an emphasis on taking the battle to the malicious hackers, something Lyons said was a vital approach.

“America’s posture in cyber defense has been largely a defensive posture,” he said. “That’s a losing strategy for America. America has to dominate the full spectrum of cyber operations.”

The post Pentagon cyber official calls advanced AI ‘revolutionary warfare’ appeared first on CyberScoop.

Two of the most advanced artificial intelligence models — Anthropic’s Claude Mythos Preview and OpenAI’s GPT-5.5 — have significantly surpassed the already-accelerating pace at which AI systems are completing autonomous cybersecurity tasks, according to separate findings published Wednesday by the United Kingdom’s AI Security Institute (AISI) and Palo Alto Networks.

The AISI, which conducts pre-deployment evaluations of frontier AI models on behalf of the British government, said both Claude Mythos Preview and GPT-5.5 have substantially exceeded the doubling trend the institute had been tracking since late 2024. Whether the results represent an isolated capability jump or the start of a new, faster trajectory remains unclear.

The AISI estimated earlier this year that frontier models’ 80% reliability cyber time horizon — a measure of how long a task takes a human expert, used as a proxy for AI autonomy — had been doubling approximately every five months. That was itself roughly half the eight-month doubling time the institute estimated in November 2025. Now Mythos Preview and GPT-5.5 have since outperformed any trend lines the institute has measured.

“Frontier AI’s autonomous cyber and software capability is advancing quickly: the length of cyber tasks that frontier models can complete autonomously has doubled on the order of months, not years,” the AISI wrote.

The clearest evidence of the capability jump came from the AISI’s cyber ranges, its structured simulations of multi-stage attacks against small, undefended enterprise networks. A newer checkpoint of Claude Mythos Preview became the first model to complete both of the institute’s ranges. It solved “The Last Ones,” a 32-step simulated corporate network attack, in 6 of 10 attempts, and completed “Cooling Tower” — previously unsolved by any model — in 3 of 10 attempts. GPT-5.5 solved “The Last Ones” in 3 of 10 attempts.

Palo Alto Networks reached similar conclusions through its own testing. The company said it began testing Claude Mythos in April as a launch partner for Anthropic’s Project Glasswing, and has since tested Claude Opus 4.7 and OpenAI’s GPT-5.5-Cyber as part of OpenAI‘s Trusted Access for Cyber program.

“The latest models are extraordinarily capable at finding vulnerabilities and changing them into critical exploit paths in near-real-time,” Palo Alto Networks wrote.

The company released security advisories covering 26 CVEs representing 75 issues — compared to a typical monthly volume of fewer than five CVEs — that were identified through AI model scanning across more than 130 products. All important vulnerabilities in its SaaS products had been patched, with patches available for all customer-operated products.

The AISI was careful to note the limits of its data. The estimates are based on a relatively small number of models, and the hardest tasks in the test suite have the least amount of human comparison data. Even so, the institute said the overall trend holds up: dropping any single model from the analysis barely moves the needle, shifting the estimated doubling time by less than a month in either direction. Separate research from METR, a nonprofit that tracks how quickly AI handles software tasks, arrived at a nearly identical figure — a doubling time of approximately four months since late 2024.

“No single benchmark result should be read as a precise measure of AI capability,” the AISI wrote. “Regardless, the direction of change and rapid growth have been consistent across the models, methodological choices and independent data we examined.”

Palo Alto Networks outlined four immediate priorities for enterprises as these models continue to grow in usage: First, find and fix vulnerabilities in code and applications before attackers do. Second, shrink the attack surface and use AI to spot security misconfigurations. Third, deploy detection and response tools across all systems, using machine learning to catch threats in real time. Fourth, build security operations fast enough to respond in minutes, because AI-powered attacks may soon unfold that quickly.

The AISI said it is developing more demanding evaluations, including new cyber ranges and the addition of active cyber defenses, to better reflect real-world conditions as model capabilities continue to advance.

The post Researchers say AI just broke every benchmark for autonomous cyber capability appeared first on CyberScoop.

The House Homeland Security Committee is digging into Anthropic’s AI model Mythos in a series of briefings and hearings, as questions proliferate on whether and how the federal government will make use of the technology touted for its ability to autonomously uncover cyber vulnerabilities.

Wednesday brought a closed-door briefing for the House Homeland Security Committee from Anthropic. The chairman of the panel’s cybersecurity subcommittee said he is planning to hold a hearing on the topic. And committee Democrats are requesting a classified briefing with Anthropic.

A committee aide who attended the briefing said it included a live demonstration of Mythos, “allowing members to see firsthand how advanced AI can identify and reason through software vulnerabilities. What we saw reinforced the urgency of ensuring that federal agencies, including our civilian cyber defenders, can responsibly access and deploy the most advanced U.S. models to find and patch vulnerabilities before foreign adversaries or criminal actors exploit them.”

A number of key lawmakers, including top committee Democrat Bennie Thompson of Mississippi and GOP cyber subcommittee chair Andy Ogles of Tennessee, told CyberScoop they weren’t able to attend Wednesday’s briefing. A second source who attended said it was a “productive” meeting.

“Members on both sides were focused on preserving U.S. advantage in AI, which basically came down to preserving our edge on compute power,” the source said. “They were also asking questions about whether the federal government was using Mythos, including about where CISA is and the impact of the supply chain risk designation.”

The Hill reported that Wednesday’s briefing was led on the Anthropic side by Logan Graham, from the company’s frontier red team, and Josh Tilstra, from the firm’s national security programs and policy team. It follows another recent closed briefing with Anthropic and OpenAI for the House Homeland Security Committee.

Ogles told CyberScoop he plans to hold a hearing of his subcommittee related to Mythos, but wasn’t able to attend Wednesday’s briefing due to scheduling conflicts. The top Democrat on Ogle’s subcommittee, Delia Ramirez of Illinois, also was unable to join due to prior commitments, but she was set to receive a rundown from staff about Wednesday’s briefing, her office said.

There’s a divide on which federal agencies are using Mythos thus far. For example: CISA reportedly isn’t, but the National Security Agency is. 

The federal divide on its use follows a Department of Defense blacklist that labeled the company a “supply chain risk” after Anthropic resisted pressure from the Pentagon to use its Claude AI model in ways the company opposed. The department says it has been using Mythos to identify cyber vulnerabilities despite the blacklist.

A turf battle is brewing within the Trump administration over testing of AI models, The Washington Post reported this week. Connecticut Rep. Jim Himes, the top Democrat on the House Intelligence Committee, said this week that it would be ‘insane” for U.S. spy agencies not to have early access to advanced AI models.

The Mythos briefing came one day after OpenAI announced its own cybersecurity initiative.

The committee aide said that “as the PRC aggressively works to close the AI innovation gap with the United States, the committee remains focused on ensuring that America’s AI leadership translates into a durable national security advantage, not a temporary lead that adversaries can copy, steal, or rapidly commoditize.”

Updated 5/13/26: to include comment from a committee aide who attended the briefing.

The post Closed briefing sets stage for House hearing on Anthropic’s Mythos and cyber risks appeared first on CyberScoop.

OpenAI has unveiled Daybreak, a cybersecurity initiative that combines the company’s large language models with its Codex agentic framework to help organizations identify, patch, and validate software vulnerabilities across the development lifecycle.

The platform is built around three model tiers: GPT-5.5 for general-purpose use, GPT-5.5 with Trusted Access for Cyber for verified defensive security workflows, and GPT-5.5-Cyber, a more permissive variant intended for specialized use cases such as authorized red-teaming and penetration testing. Each tier carries different safeguard levels and access controls, with the most capable tier paired with stronger identity verification and account-level oversight.

“For cyber defense, it means seeing risk earlier, acting sooner, and helping make software resilient by design,” a company blog post reads. 

OpenAI did not respond to CyberScoop’s request for further comment. 

Daybreak arrives weeks after Anthropic unveiled Project Glasswing, built around Claude Mythos Preview, a cybersecurity-focused AI system Anthropic has described as capable of autonomously identifying software vulnerabilities at scale. Anthropic has kept access to Mythos tightly restricted, citing both safety concerns and national security considerations, and has not made the model commercially available.

A tiered approach to access

The structure of Daybreak reflects a deliberate effort to calibrate access against the risk these models present. The standard GPT-5.5 model is available for general enterprise and developer work. GPT-5.5 with Trusted Access for Cyber is aimed at security professionals engaged in defensive workflows, including vulnerability triage, malware analysis, detection engineering, and patch validation. GPT-5.5-Cyber, the highest-capability tier, is currently in preview and reserved for specialized workflows under controlled conditions.

OpenAI has framed the access controls as a response to the dual-use nature of the underlying technology. The same AI capabilities that allow defenders to understand relationships across codebases, identify subtle vulnerabilities, and accelerate remediation could be misused, the company acknowledged. The platform pairs expanded capability with what OpenAI describes as trust, verification, proportional safeguards, and accountability.

“We don’t think it’s practical or appropriate to centrally decide who gets to defend themselves,” the company said in a prior blog post related to the Trusted Access for Cyber program. “Instead, we aim to enable as many legitimate defenders as possible, with access grounded in verification, trust signals, and accountability.”

Industry partners and government discussions

Several major technology and cybersecurity companies are already working within the Trusted Access for Cyber framework, including Cisco, Oracle, CrowdStrike, Palo Alto Networks, Cloudflare, Fortinet, Akamai, and Zscaler.

Anthony Grieco, Cisco’s chief security and trust officer, described the technology as a “force multiplier for defenders,” noting that models like GPT-5.5 are changing the pace of security operations, from incident investigation to proactive exposure reduction. He added that the value of the technology lies not in the model alone but in the enterprise framework built around it.

At the federal level, the Trump administration is weighing how Anthropic’s Mythos will be used to protect government networks, with Federal CIO Greg Barbaccia telling CyberScoop last month he sees its potential to strengthen federal cyber defenses and the significant uncertainties that remain about how it would perform in real-world conditions.

Elsewhere, the European Commission is in discussions with OpenAI about potential access to its advanced AI models for identifying cybersecurity vulnerabilities. 

Other industry experts told CyberScoop that while these models are very good at finding vulnerabilities, that’s only part of the puzzle when it comes to an enterprise security plan. 

“The question that determines breach impact is not how fast you find the vulnerability. It’s how far a compromised identity can move before anyone knows it’s compromised,” said Doug Merritt, chairman & CEO, Aviatrix, a cloud security company. “That’s an infrastructure problem — what is each workload allowed to reach, on every path, independent of whether the breach has been detected? No patching tool answers that. Containment does.”

Jared Atkinson, CTO of SpecterOps, an identity management company, says defenders need to focus on what attackers can reach once inside, while still working to identify vulnerabilities faster.

“AI will accelerate portions of offensive security operations, but it does not fundamentally change the underlying problem defenders face. Most organizations still struggle to see and manage the attack paths that connect initial access to critical systems and data,” he said. “As these tools mature, visibility into identity exposure and post-compromise attack paths becomes increasingly urgent.”

A widening competition

The competitive cybersecurity dynamic between Anthropic and OpenAI has been building for months. OpenAI publicly announced the Trusted Access for Cyber program before Anthropic’s Glasswing rollout and has since expanded it to thousands of individuals and organizations. In April, the company released GPT-5.4 Cyber, a model variant specifically optimized for cybersecurity tasks, including testing and vulnerability research, governed by Know-Your-Customer and identity verification requirements.

Cybersecurity experts in the United States and United Kingdom have described Claude Mythos as a meaningful improvement over previous frontier models in identifying cybersecurity vulnerabilities, though debate continues over its practical impact on information security. GPT-5.4 Cyber has similarly been fine-tuned for testing and vulnerability research, with OpenAI indicating it intends to make iterative improvements as the program matures.

OpenAI’s stated intent is to expand access to Daybreak’s most capable models over time, working alongside industry and government partners as it deploys what it describes as “increasingly more cyber-capable models” through an iterative deployment approach. The company has indicated it is cautious about exercising too much centralized control over which sectors or industries participate in the program.

CEO Sam Altman framed the initiative in broad terms. “AI is already good and about to get super good at cybersecurity,” he wrote on X. “We’d like to start working with as many companies as possible now to help them continuously secure themselves.”

The post Daybreak is OpenAI’s answer to the AI arms race in cybersecurity appeared first on CyberScoop.