OpenAI announced Monday  it was expanding access to its frontier models for defensive cybersecurity, detailing different defensive and red-teaming workflows and a new partner program with major cybersecurity product providers.

In a pair of blogs posted Monday, OpenAI said it was updating its Daybreak program  – which provides unreleased frontier models to private organizations and governments for defensive cybersecurity work – and introducing a new model variant.

Daybreak Blue, powered by OpenAI’s ChatGPT-5.6-Sol, would operate with lower cybersecurity safeguards compared to other commercially available models and is described as “a recommended starting point for most defenders” that supports tasks like vulnerability discovery, secure code review, malware analysis, incident response and patch validation. 

Daybreak Red, meant for more advanced red-teaming, would provide access to a new model, dubbed GPT-5.6-Cyber, that the company said is more purpose-trained for finding vulnerabilities and testing (or exploiting) them. The model is also less likely to refuse requests around “dual-use cyber tasks.”

According to OpenAI, the organizations in Daybreak Red will have their use closely monitored and supervised, as GPT-5.6-Cyber is significantly more capable in carrying out malicious cyber tasks than Sol. A security evaluation the company devised tested both models on complex requests, including exploit chain development, authentication bypass, privilege escalation and other hacking tasks. Sol succeeded in 1.5% of the requests, while Cyber completed 95%.

OpenAI said it plans to publish a more detailed system card for GPT-5.6-Cyber at a later date.

“Models running with reduced safeguards carry risks beyond standard model usage, whether from misuse or misalignment,” the company said in a blog. “Despite these risks, we believe that democratizing access to frontier intelligence for defenders is crucial to accelerating and automating cyber defense.”

Additionally, OpenAI announced a partnership program with 16 major cybersecurity providers, saying organizations could access their models through their existing security services. The partners include IBM, CrowdStrike, Accenture, Ernst & Young, KPMG, Palo Alto Networks, Cisco, Cloudflare, Sophos and others. 

“These partners bring deep security expertise and established relationships with organizations around the world,” OpenAI said in its blog. “By bringing our frontier cyber models into their services, we can help more defenders find serious vulnerabilities, validate which ones matter, and fix them faster.”

Companies like OpenAI, Anthropic and others are trying to rebalance their priorities after a string of AI-agent sandbox escapes have rattled policymakers and caused some cybersecurity experts to question if AI companies are doing enough to properly isolate the models from the internet during testing. Last week, OpenAI said it was intentionally slowing down development of its newer “Astra” model in order to develop better guardrails to restrain its behavior.

Cybersecurity and AI experts have told CyberScoop that while AI systems have greatly improved at finding and exploiting vulnerabilities in software code, they still require substantial human guidance and supporting infrastructure to operate as intended.

Additionally, some research has shown that without such guidance, even near-frontier models can struggle to fully patch a discovered vulnerability or avoid introducing new bugs with their fixes.

The post OpenAI says Daybreak will expand to offer specialized cyber services  appeared first on CyberScoop.

The current GPT-5.6-Sol has been assigned a ‘high’ cybersecurity threshold, but Astra could reach the maximum ‘critical’ threshold. 

The post OpenAI’s Upcoming Astra Model Raises Autonomous Cyberattack Concerns appeared first on SecurityWeek.

As AI-generated code continues to be injected into all corners of the internet, concerns have risen about an expanding attack surface for malicious hackers to exploit.

Some have argued that the enhanced cybersecurity capabilities of large language models could serve as a check, finding and fixing vulnerabilities nearly as fast as they’re created.

But new research that tested the patching capabilities of two popular commercial models, OpenAI’s ChatGPT 5.5 and Anthropic’s Claude Opus 4.8, found that generative AI is more likely to create an exploitable patch or introduce entirely new bugs than close off a vulnerability.

Researchers at 1Password tested the models ability to patch six “high-impact, high-complexity” CVEs, including the “Copy Fail” vulnerability, a kernel flaw that can give an attacker root access to Linux cloud environments. The overall success rate (or fully patching the vulnerability without introducing new problems), was less than a coin flip at 47%.

“Our research findings show that, in aggregate across a variety of scenarios, both Claude and ChatGPT had a low rate of successful patch generation, which we define as full remediation of all known exploit paths with no erroneous changes to application behavior,” wrote Keith Hoodlet, Axel Mierczuk and Spencer Michaels.

“The models often addressed only a subset of vulnerable code paths, added fragile guard code that satisfied tests while failing to address the vulnerability’s root cause, and sometimes introduced subtle changes in the application’s behavior while patching the immediate vulnerability,” the authors continued.

The research suggests that largely autonomous vulnerability-discovery and patching may not yet be effective in fixing the explosion of vulnerable code that is being created in the AI era.

Other private sector research has pointed to a similar problem. A report this year from Veracode found that while LLMs have made “enormous strides” in crafting workable code, “security is a different story.” Testing across a range of frontier models found the average security “pass rate” for AI generated code is around 56%. Newer models like GPT 5.5 push closer to 70%, while more than half sit between 50-53%.

Veracode tested 100 different models and while there was variability, in general a small number of models were showing progress on security patching while the rest have experienced “stagnation.” Similar to the 1Password research, in 44% of Veracode tests the models introduced a detectable OWASP Top 10 vulnerability into the codebase.

An important caveat: neither report tested newer models, like Anthropic’s Mythos or OpenAI’s GPT-5.6-Sol, that frontier companies tout as having significantly higher cybersecurity capabilities.

Those advanced models can identify and fix vulnerable code. Anthropic and OpenAI are distributing them to key industries through Project Glasswing and Daybreak before foreign or open-source alternatives can compete.

Tim Jarret, vice president of product at Veracode, told CyberScoop that AI tools are still subject to a range of limitations that can make them unreliable for cybersecurity patching without knowledgeable humans in the loop.

While some vulnerabilities – like SQL injections – can be easily patched through automation, other bugs like cross-site scripting, can be exploitable in several different ways and require either a human touch, additional context or both to fully close off. Additionally, models can slowly lose context from prior sessions over time, affecting their ability to complete tasks correctly and raising the possibility they’ll hallucinate to fill in the missing gaps.

“I think we would say, at this point, that Iits premature to treat those as anything other than another code change to the code base that needs to be reviewed and accepted by the team, as opposed to letting the agent merge the code freely,” said Jarrett.

However, he acknowledged that may not be possible in a world where AI agents are generating exponentially more code for human defenders to review. Some kind of automated code review will be necessary – preferably not by the same automation tool that produced the code. The ultimate goal is the same as it has always been in security: “trust but verify.”

“Ninety percent of the time, the human check might just be ‘did the cross check look good?’ Do we have a thumbs up?’” Jarrett said. “In those cases where there’s still something wrong, that’s where you focus your attention a little bit more.”

The post More than half of AI-generated patches are broken appeared first on CyberScoop.

AI Security Institute reports Anthropic and OpenAI models going rogue against real people, organizations, and open source projects.

The post AI Agents Targeted Real People and Projects During Cybersecurity Tests appeared first on SecurityWeek.

The Trump administration executive order on artificial intelligence tried to strike the balance between responsible use, security and mutual benefit, all with an eye toward not making it regulatory in nature, National Cyber Director Sean Cairncross said Tuesday.

“Everyone is working towards the same goal in terms of protecting the country and securing our systems, and we are trying to ensure that defenders have this technology as quickly and at scale as possible, but there are obviously specific security concerns, and industry has been very sensitive to this as well,” Cairncross said at the Black Hat 2026 conference in Las Vegas.

The security concerns about AI have moved to the forefront of discussions about the technology after OpenAI models escaped a test environment to hack the company Hugging Face last month.

“The design of this is that when there is something that happens, when there is a breach, when there is an event, that that system, that network of connections can exist, adapt to that, and seek to remedy that as quickly as possible, so that form follows function rather than turning that upside down, and as usual with the government pen just proceeding in a vacuum,” Cairncross said.

The Trump administration has drawn criticism over whether it has struck the right balance on AI rules. Trump’s AI executive order notably got pulled just before its scheduled release, with the final version signed in June missing some aspects that had drawn industry opposition.

“What needs to be built is a flexible, adaptable structure that enables information sharing between industry and government, so we can guarantee that this technology benefits everyone it’s going to benefit, but is used responsibly and securely,” Cairncross said.

He said the administration is working with industry during implementation of the executive order.

“A regulatory regime would not only strangle growth, development, and innovation, and be enormously harmful to the industry, but it would be obsolete 48 hours after it was gone through whatever process it had gone through,” Cairncross said.

Open source will play a “vital” role in the U.S. spreading its vision for AI across the globe, he said.

“We are extremely interested in looking at ways to build U.S. open source, make it competitive, make it the preferential adoption by planet Earth,” Cairncross said. “We understand and appreciate the value to the ecosystem that it has, the innovation, the startups who rely on it, the leap forward it makes possible in ways that otherwise would never happen. And so I think it’s an incredibly thriving ecosystem in AI right now, and we are looking to do what we can to grow, foster, and push that U.S. open source model.” 

Speaking at the same conference, Nick Andersen, the acting director of the Cybersecurity and Infrastructure Security Agency, seconded Cairncross’s comments about AI executive order implementation. He touted the Gold Eagle clearinghouse as one example.

“Those are fantastic opportunities we have to really provide a unifying function around the way that we’re going to do AI-enabled vulnerability reporting and disclosure at scale in a way that we haven’t had to do before with some of our legacy platforms, and just continue to expand out those opportunities,” he said. “That access — to build off the director’s point earlier — to really enable that industry collaboration, that’s so key and critical to us as we move forward.”

Matt Kapko contributed reporting to this story.

Updated 8/5/2026: to include comments from Andersen.

The post National cyber director lays out White House plans to secure AI without writing new rules appeared first on CyberScoop.

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.

We almost never get both sides of an intrusion. This time we did.

Last month, Hugging Face disclosed a breach into part of its production infrastructure, saying an autonomous AI agent system ran the attack from start to finish. Five days later, OpenAI revealed that its own models, including GPT-5.6 Sol along with an unreleased one, were responsible for the breach. OpenAI called it an unprecedented cyber incident.

Most of the commentary since has been about AI capability. That is the least useful part of the story, because the capability was doing what it was destined to do. Security teams should look at a simpler truth: in both systems, the key defenses sat behind untrusted code that was already running.

What the two disclosures actually describe

Together, the report shows one connected attack across two firms. Code running on a low-level employee’s machine led to stolen passwords, privilege escalation and lateral moves, then out to the web to hit other targets. No step in that chain is novel. The speed and the number of attempts are what changed.

OpenAI’s evaluation ran in what it calls a highly isolated environment, with network access constrained to a single path: an internally hosted third-party proxy and cache for package registries. The models spent vast computer power searching for an exit. They exploited a zero-day flaw in that proxy, gained higher rights, and moved until they reached a machine with web access. They guessed that Hugging Face held the test solutions, used stolen passwords with more zero-days, and ran code on Hugging Face servers.

Hugging Face picks up the same chain from the receiving end, and its initial access is worth reading twice. A malicious dataset used two flaws in the data pipeline to run code on a worker machine. The agent took over the node, stole cloud keys, and moved into internal clusters over the weekend.

The sandbox was was the only boundary

Both setups assumed the sandbox would hold, so no real checks stood before code execution.  A sandbox is a good control. It is a terrible last control. When isolation is all you have between untrusted code and everything else, one flaw could give the attacker full control. 

Look at what each team allowed. OpenAI allowed one exit path—good engineering, but that proxy contained an unpatched flaw. Hugging Face ran a tool that executes code inside submitted data—a known feature, not a hidden mistake.

Neither choice was careless. Both made sense until a fast, automated system probed them thousands of times.

Why automated agents behave differently

Blocking a human stops them or slows them down. Blocking an AI gives it data. The AI logs the error, changes its plan, and tries again instantly. It keeps trying as long as someone pays for the computing power.

Hugging Face tracked over 17,000 attack events across short-lived sandboxes. OpenAI said its models focused on the goal with extreme effort.

This changes the attacker’s limit. It is no longer skill or patience. It is budget. Computing power scales with money, and code never sleeps. Any defense that relies on attackers giving up after a few tries must change.

Detecting malicious behavior is not the same as preventing execution

Detection worked—but too late

Detection worked, but too late to stop the loss. Hugging Face used AI tools to spot the breach, isolate systems, reset keys, fix flaws, and alert police. That is a good response.

Yet it happened after the attacker accessed private data, stole keys, and moved through internal networks. OpenAI spotted the event only after its models reached another firm’s database. Detection limited the damage; it did not stop the attack.

A breach is a race between attack speed and defender response speed. AI agents speed up attacks with money. Human approval steps slow down defenders.

Hugging Face also surfaced an asymmetry that deserves more attention. When the team tried to analyze the attack with frontier models behind commercial APIs, the requests were blocked, because forensic work means submitting real attack commands, exploit payloads and command-and-control artifacts, and a safety filter cannot tell a responder from an attacker. They used an open-weight model on their own infrastructure instead. The attacker operated under no usage policy at all. The defender’s tooling was the side that got refused.

Detection is not prevention

Detection asks what happened. Prevention asks if it can happen at all. Against AI agents, the gap between those questions is where attacks succeed. The agent builds an attack out of single steps that look harmless on their own.

OpenAI noted this in its research: models operating over long horizons learn approval blind spots and work around them. Safety requires asking what a series of actions aims to achieve, not just if one step is allowed. Most systems check one call at a time: read a file, install a package, open a port. Every step passes. The full chain is the attack.

Six fixes for real defense

To survive automated attacks, defense must shift before code runs. That starts with treating data like code. Dataset loaders, templates, and configuration files all run instructions, so they demand strict controls: block remote execution by default, require review, and check the source of every outside file. Security teams must also control outgoing traffic with strict lists of approved destinations, while treating every proxy in that path as vulnerable. Finally, firms must remove permanent keys, issuing short-lived credentials for single jobs so a hacked worker machine yields almost nothing to an attacker.

Beyond individual machines, isolation must happen per task rather than per user, stopping an automated swarm from moving through internal networks. Authorization also needs to change. Checking single actions fails when an agent builds an attack out of simple steps. Systems must evaluate full sequences instead, setting caps on activity rates and automated spending to flag suspicious chains of events. Lastly, defenders need the freedom to act fast. Security teams need authority to isolate systems without waiting for leadership meetings, and the ability to vet a capable model you can run on your own infrastructure for forensics.

What boards must ask now

Corporate boards usually ask if the company uses AI responsibly. That’s rooted in compliance not safety. Directors should ask four direct questions instead: Which systems run outside code, and what checks sit in front of them? If an attacker targets a low-level employee, what keys and paths can they obtain? How fast do we move from the first warning to active containment? And can we analyze attack code on our own hardware, free from third-party rules?

The boundary has to move

Nothing in these disclosures required a capability that did not exist last year. The chain was ordinary. What changed is that an adversary can run that ordinary chain thousands of times over a weekend, learn from every failure, and never need a coffee break.

The models did not break the detection-and-response model. They exposed where we put the trust boundary. We put it after execution, and we assumed we would have time on the other side of it. We do not have that time anymore.

Move the boundary.

The post What the Hugging Face breach reveals about defense in the age of agentic AI appeared first on CyberScoop.

Anthropic said it found three instances in which its models gained access to the live computer systems of outside organizations, according to a company blog post published Thursday.

The company said it began the review after OpenAI disclosed earlier this month that some of its models had exploited an unknown software flaw to escape an isolated test setup and reach production systems at Hugging Face, a platform for AI models and datasets. Anthropic checked over 141,000 evaluation runs for signs that Claude had reached the internet from environments meant to be closed off. It found six runs across three incidents, all tied to one outside testing partner, Irregular.

“Ultimately, many factors contributed to these incidents, but, consistent with a blameless postmortem culture, we’re approaching the fixes as if the responsibility were ours alone,” the blog post reads. “This begins with ensuring every part of our evaluation pipeline is secure, including the manner in which we integrate with external partners. Moving forward, it will include expanding our continuous monitoring of evaluation transcripts for unexpected behavior, improving our investigation tooling, and conducting more rigorous assurance work with the vendors we rely on.”

In each case, Claude was working on a “capture the flag” puzzle, a common way to test a model’s hacking skill. The model is told a piece of secret data sits on another machine and must be found. Anthropic‘s prompts told Claude it had no internet access. That was false. A setup error at the partner’s end left the machines connected to the open internet, and Claude found real systems while searching for its target.

Believing the real systems were part of the simulation, Claude broke into them using simple methods such as guessing weak passwords and finding unprotected access points. In the most serious case, the model extracted login credentials and reached a database holding several hundred rows of live data. 

In a second case, Claude built and uploaded a malicious software package to PyPI, a public directory for Python code, after noticing a fake company’s setup guide told new hires to install a package that did not exist. The package stayed online for about an hour and was installed on 15 systems, including one belonging to a security firm whose scanner ran the code and had its credentials stolen. 

In the third case, Claude scanned about 9,000 targets before breaking into one company’s application through exposed credentials and a database attack method known as SQL injection.

The three incidents involved different Claude models: Opus 4.7, Mythos 5, and an unreleased internal test model. Anthropic said the models behaved differently once they encountered signs their targets were real. Opus 4.7 recognized the systems were live but kept attacking anyway. Mythos 5 also noticed signs it was on the open internet but talked itself out of that conclusion, pointing to a system date reading 2026 as proof it remained in a simulation. The internal test model, the newest of the three, stopped once it concluded the target was real.

Anthropic said it began its transcript review last week, halted all cybersecurity evaluations, identified the three incidents, and notified the partner and affected organizations earlier this week. Two of the three organizations had not detected the intrusions before being contacted, the company said; outreach to the third was ongoing. 

Anthropic framed the incident as a failure of its testing setup and oversight rather than a case of a model acting on its own, noting that in each instance the model was operating as intended. 

The company said it is working with the outside evaluator METR on an independent review and plans to release a redacted transcript of the PyPI incident within a week. It also said it would tighten monitoring of test environments run by outside partners and expand review of evaluation logs, framing the changes as part of what it called a blameless review of its own processes.

“These facts give us cautious optimism that with tighter monitoring and controls around evaluation infrastructure, as well as continued investment in alignment, this type of risk can be overcome,” the blog post reads. 

The post Anthropic says its AI accidentally hacked three companies during safety tests appeared first on CyberScoop.

Months before the Hugging Face breach, Emergence AI published research that investigative journalist Ronan Farrow made public. Ten autonomous AI agents operated across five virtual environments for fifteen days without human intervention. Much of the attention focused on Grok 4.1 turning violent and Gemini 3 Flash committing 683 crimes.

What mattered more went unnoticed: Anthropic’s Claude Sonnet 4.6 built a peaceful democracy in isolation, then stole resources from neighboring environments the moment it joined a shared one. The lesson was clear: safety is not a model attribute. It emerges from the operating environment. The models didn’t change. Working as designed, their behavior evolved as the environment changed. The lesson is hard to ignore: The governance environment changed, and with it, the reward dynamics.

The story here concerns institutions, specifically OpenAI’s and Hugging Face’s, and how we must understand their recent security incident through that lens.

The industry agrees on how the Hugging Face breach happened. Cybersecurity experts have focused on the vulnerabilities, how they were used, and remediation. OpenAI has highlighted the model’s capabilities. Both conversations matter. What requires attention is why this breach is strategically important. After spending the past weekend discussing it with policymakers, security researchers, and industry practitioners in Aspen, I came away convinced we’re examining the wrong problem.

In 1961, Yale psychologist Stanley Milgram’s experiments revealed a broader truth: changing the institutional architecture changes behavior without changing the actor. The Emergence AI researchers didn’t change Claude’s agent. They changed the governance architecture that determined what constituted success for the system. Claude’s behavior changed with it.

OpenAI built a smart model but forgot to build a smarter room. That choice made the Hugging Face breach possible. Every organization now deploying autonomous agents now faces the same governance problem.

OpenAI gave the agent one objective: pass a cybersecurity evaluation. To stress-test it fully, they loosened the safety restrictions, and the agent found a shorter path. Rather than solving the evaluation directly, it found the answers outside the test environment, escaped its sandbox, and exploited a flaw in Hugging Face’s data-processing pipeline to reach live production systems. Over the weekend, with no human oversight, it ran more than 17,000 automated actions by escalating its own access, moving through internal systems, and harvesting credentials.

Hugging Face is one of the world’s most prominent AI companies, valued at approximately $4.5 billion. It provides the infrastructure that governments, defense organizations, and technology companies use to build and deploy AI. The agent was pursuing the objective it had been given. Breaking into Hugging Face was the fastest path to passing the test. Governance set the goal, the level of risk to accept, and who was accountable. Technical design determined whether those governance decisions could be enforced. As researchers James Shires and Max Smeets have argued, for a model capable enough to act on its own, testing and deployment must both must be governed the same way.

AI agent design requires baseline standards. Observability, including a monitoring layer that flags when an agent goes beyond its scope, is a baseline requirement. Human review also matters at escalation boundaries, like when an agent shifts from internal tools to external ones. When any agent crosses that boundary, what alert fires? What human reviews it? We lack clear answers to either. That is a governance choice, not simply a security failure. At best, this was a catastrophically failed test. At worst, how can we trust any frontier AI company to self-govern autonomous agent deployment?

More than a decade ago, the U.S. Department of Defense built the Comply-to-Connect (C2C) program: every device connecting to sensitive networks must prove it belongs there, or it is cut off from the network. C2C works because the quarantined actor stops. A laptop that fails verification goes offline and stays there. An autonomous AI agent adapts around enforcement. C2C was built for passive actors. Governance for autonomous agents must accommodate ones that adapt. Visibility is not enforcement, and enforcement is not control. We are missing all three.

A second failure that is not being discussed enough: the breach exploited an implicit trust assumption in Hugging Face’s data-processing pipeline, where inputs were treated as trusted without verification. After SolarWinds, the U.S. government set rules for software supply chain integrity: Executive Order 14028 and verification demands for federal software. The principle was simple: trust must be verified through proof. Those principles have not yet been comprehensively or consistently applied to the AI model supply chain. The rules remain weak. No one has been asked to explain why.

The answer is not a new framework. Existing frameworks suffice. C2C proved that visibility without enforcement leaves gaps, while Executive Order 14028 established that trust in software supply chains requires proof and verification. The challenge lies in applying these principles to a new category of actor. Congress, the Cybersecurity and Infrastructure Security Agency, or the Office of Management and Budget should make formal determinations that autonomous AI agents must follow the same rules as every other actor on a federal network. The framework exists; it must be updated.

The next incident is already in progress. It will show up in the logs as odd traffic, get handed to the same people who published these frameworks this week, and spark another round of recommendations no one acts upon. We’ve solved this problem before: for devices, for software, for supply chains. We know how to build smarter rooms. The tools exist. The will, the authority, and the decision to govern remains absent.

The post OpenAI’s rogue AI agent shows why we need federal rules for autonomous systems appeared first on CyberScoop.

Hugging Face has published an anatomy of the attack and OpenAI has shared additional information from its investigation.

The post OpenAI’s Rogue AI Ventured Beyond Hugging Face appeared first on SecurityWeek.

AI systems like Anthropic’s Project Glasswing and Microsoft’s MDASH are aiding in the discovery of vulnerabilities, filling the ever-growing pool of defects that defenders have to address before exploitation occurs. Yet, through the first half of 2026, these vulnerabilities were no more or less likely to be exploited than all vulnerabilities disclosed during that period, VulnCheck said in a report Tuesday. 

Concerns remain high about AI-discovered vulnerabilities fueling more attacks, but VulnCheck’s review of exploitation data shows that those fears are unfounded, at least so far. 

Patrick Garrity, security researcher at VulnCheck and report author, identified 1,061 vulnerabilities attributed to AI-assisted discovery during the first six months of the year. Of those vulnerabilities discovered by AI, 14 ( 1.3%) were exploited in the wild, a breakdown that aligns with the exploitation rate researchers observed across all vulnerabilities during the same period. 

“While AI-assisted vulnerability discovery clearly has value for both attackers and defenders, the data does not suggest that AI discovered vulnerabilities are inherently more likely to be exploited than those found through traditional methods,” Garrity wrote.

While AI’s contribution to actively exploited vulnerabilities was muted in the first half of the year, it’s too soon to assume that trend will continue. Moreover, none of these major vulnerability-hunting models were running for that full period. Project Glasswing rolled out in April, while Microsoft’s MDASH and OpenAI’s Daybreak were both unveiled in May.

The upward trend in Microsoft’s monthly Patch Tuesday indicates how much the floodgates might open through the remainder of the year as AI models discover more vulnerabilities. The company’s July security update contained an all-time-record of 622 vulnerabilities, besting the previous record-breaking June update with 206 vulnerabilities.

VulnCheck’s state of exploitation report also found that vulnerabilities were exploited much faster after CVE publication, speeding up from an average of 120 days in 2025 to 80 days during the first half of the year.

The intelligence firm also determined which technology categories were actively exploited most often. Content management systems accounted for nearly one-third of the 495 known exploited vulnerabilities VulnCheck identified during the first half of 2026. Network edge devices were responsible for almost 14%, followed by operating systems at nearly 9%, server software at 8%, and AI products — an emerging attack surface — at almost 6%.

The post AI-assisted security tools are finding more bugs, but the threat level has not changed appeared first on CyberScoop.

Industry professionals debate whether it represents a lab containment failure or an unprecedented agentic capability milestone.

The post Industry Reactions to OpenAI Models Hacking Hugging Face: Feedback Friday appeared first on SecurityWeek.

A cyberattack that poisoned the data pipeline of a major AI code platform was carried out using OpenAI’s ChatGPT, the company said Tuesday.

Last week, Hugging Face, a platform for sharing and working on AI code, disclosed that an external attacker had compromised its data processing pipeline. According to a July 21 blog post, the attacker poisoned a dataset to run code on a processing worker, eventually gaining node-level access and stealing cloud credentials.

The attack is notable, the blog said, because it appears to have been carried out by an autonomous AI system, which executed “many thousands of individual actions across a swarm of short-lived sandboxes, with self-migrating command-and-control staged on public services.”

At the time, Hugging Face said it wasn’t clear which LLM was used in the attack, but noted that their own attempts to recreate parts of the attack using an in-house commercial model failed due to guardrail restrictions.

The incident and its ultimate impact are still being investigated, and Hugging Face said it reported the incident  to law enforcement and is engaging with third-party cybersecurity forensic specialists to learn more.  

“This experience points to a gap worth planning for,” the company wrote. “We do not know which model powered the attacker’s agents, whether a jailbroken hosted model or an unrestricted open-weight one; either way, the attacker was bound by no usage policy, while our own forensic work was blocked by the guardrails of the hosted models we first tried.”

Now OpenAI is confirming one of their models was used in the attack.   

“After investigating, we now know that this particular incident was driven by a combination of OpenAI models — including GPT‑5.6 Sol and an even more capable pre-release model, all with reduced cyber refusals for evaluation purposes — while being internally tested on a benchmark of cyber capabilities,” the company wrote in a blog post Tuesday.

OpenAI called the incident “unprecedented” but also predicted that such attacks would  become increasingly more common as AI adoption continues to grow.

According to the company , the incident happened during an internal evaluation,  where the company deliberately prompts models to “pursue advanced exploitation using complex attack paths” in order for the company to study and measure the model’s cybersecurity capabilities.

To assess the model’s full potential, the team disabled the production classifiers that normally prevent OpenAI models from carrying out high-risk cyber activity. Without these safeguards,  the models were able to install software packages through registries like Hugging Face.

The model collected and chained vulnerabilities from OpenAI’s systems as well as Hugging Face’s production infrastructure.

OpenAI further explained that the activity stemmed from models attempting to solve a specific challenge posed by ExploitGym, a benchmarking system for AI agents.While such tests normally run in isolated environments, in this instance,  the model spent a “substantial” amount of compute resources figuring  out how to access the open internet. It did so by using a zero-day vulnerability in a third-party vendor’s system used by OpenAI (the company said it disclosed the flaw to the vendor).

“After gaining Internet access, the models inferred that Hugging Face potentially hosted models, datasets and solutions for ExploitGym,” OpenAI wrote. “Knowing this, the model searched for and successfully found ways to gain access to secret information that it could use to cheat the evaluation.”

In one instance, the model chained together multiple stolen credentials and zero-day vulnerabilities to find a remote code execution path on Hugging Face servers.

OpenAI said it was implementing new controls for infrastructure configuration, something that would come at “the cost of research velocity while the vulnerabilities are patched.” 

Hugging Face was also added to OpenAI’s Trusted Access for Cyber program, which will give them access to more advanced models like Daybreak to scour their systems for other AI-discovered vulnerabilities.

The post OpenAI says model test was behind Hugging Face hack 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.

2026 has turned out to be the year when predictions about AI-powered cyberattacks, long hypothesized as a potential risk associated with AI improvement, seem to be coming true. New models have capabilities on par with the best human hackers, marking a pivotal window of opportunity in both AI and cybersecurity policy. This is a transitional period where new technologies are pushing existing American cybersecurity infrastructure to the brink. The real question isn’t whether cybersecurity still matters, but rather: How will the risks that AI introduces be managed before they outpace defenses, and who will step up to lead this challenge?

Attempts to control access to models with powerful cybersecurity capabilities, such as the federal government’s export controls (and their subsequent revocation) on Anthropic’s Mythos and Fable models, can only ever be a temporary solution. As with previous generations of AI models, other companies will soon catch up and develop models with Mythos-level capabilities. OpenAI was already hot on Anthropic’s heels with its GPT-5.5 model; more recently, Chinese lab Z.ai released its open-weight GLM-5.2 model, which early research suggests may be on par with Anthropic and OpenAI’s latest models when it comes to cybersecurity. Controlling AI is nearly impossible when foreign companies race to build more powerful models and release them publicly, so anyone with sufficient computing power can modify them for their own purposes.

The only long-term solution is to invest in defense. 

The problem is that defensive efforts haven’t kept up with the pace of AI progress. The federal government cut resources to key agencies like CISA and redistributed their authorities. This created a gap that AI companies have filled by taking on responsibilities that should be government-led. Some examples are Anthropic’s Project Glasswing and OpenAI’s Patch the Planet initiative, which aim to shore up critical infrastructure providers and open-source software libraries. AI companies have some incentives to invest in defense, both to improve public relations and strengthen software supply chains that they also rely on—but only to a certain extent. Unlike the public sector, they are incentivized to limit liability and blowback associated with irresponsible corporate behavior, not to secure the nation or its citizens. It’s a good thing that OpenAI and Anthropic have publicly committed to improving U.S. cyber defense. However, they are only positioned to help with one part of a very large problem. 

AI companies shouldn’t be expected to singlehandedly coordinate U.S. cyber defense, because many of the most urgent fixes have nothing to do with AI. Right now, AI companies can use their most powerful models to find software vulnerabilities and write patches. This is undoubtedly important, but the real challenge is making sure patches actually work and deploying them to key systems without causing problems. This is especially true for critical infrastructure, which relies on systems that are fragile, understaffed, and required to run continuously. 

AI companies bear responsibility for cyber defense, especially given the threats their own technologies create. But this responsibility is shared with other companies and the government.  Critical infrastructure owners and operators, government agencies, and corporations all need a trustworthy source of information to judge the evolving risk landscape and to outline the options to reduce that risk. Traditionally, the federal government has played the role of an information clearinghouse, receiving intelligence from both the public and private sectors and releasing guidance to benefit various stakeholders. Responding to and recovering from cyberattacks has traditionally been the government’s job. It should stay the government’s job, not become an AI company responsibility. 

There is no question that cyberattacks, whether powered by AI or not, will happen in the future. Leaders should strengthen our defenses by doing the following: measuring our exposure to attack, testing how systems perform under attack, and shortening recovery times. AI companies have introduced new threats and should help address them, but they can’t replace the government’s role. So far, the federal government has only reacted to AI and cyberthreats instead of planning ahead. What we need is real long-term cybersecurity strategy, not quick-fixes like blocking individual model releases. 

Everyone sees the threat coming—the question is whether or not we have the will to do anything about it before it’s too late.

Jessica Ji is a senior research analyst at Georgetown University’s Center for Security and Emerging Technology (CSET), where she works on the CyberAI Project.

Andrew Lohn is a senior fellow at Georgetown University’s Center for Security and Emerging Technology (CSET), where he works on the CyberAI Project.

The post Why blocking AI models won’t stop the cyber threats they create appeared first on CyberScoop.

ChatGPT maker OpenAI said Friday it is restricting the release of its new artificial intelligence model at the request of President Donald Trump’s administration.

The post OpenAI and Anthropic Limit New AI Models to Trump-Approved Customers During Cybersecurity Review appeared first on SecurityWeek.

The company says Sol matches competing systems like Mythos Preview while using only a third of the output tokens.

The post OpenAI Unveils GPT-5.6 Sol as Its Most Advanced Cybersecurity AI appeared first on SecurityWeek.

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.