Hisham Sarwar
In a stunning development that sent shockwaves through the global AI industry, Anthropic, the company behind the popular Claude AI, was forced to pull its two most advanced models, Fable 5 and Mythos 5, completely offline on Friday, June 12, 2026. The directive came directly from the United States government, citing national security concerns.
This marks the first time in history that a leading AI company has been compelled by federal authority to take publicly deployed, commercially live AI models offline.
What Exactly Happened?
At 5:21 PM Eastern Time on Friday, Anthropic received a formal letter, signed by Commerce Secretary Howard Lutnick, issuing an export control directive. The order demanded that Anthropic immediately suspend all access to Fable 5 and Mythos 5 for any foreign national, whether they were inside or outside the United States, including Anthropic’s own foreign national employees.
Because Anthropic could not technically separate access by nationality across its global user base, the company had no choice but to disable both models entirely for all customers worldwide.
“The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.” (Anthropic’s official statement)
If you visited Claude’s platform Friday evening, you would have seen a simple message: “Fable 5 is temporarily unavailable.”
Who Are Fable 5 and Mythos 5?
To understand why this matters so much, you need to know what these models are.
Mythos 5 is Anthropic’s most capable AI model ever built. The company first previewed it in early April 2026 and deliberately kept it tightly restricted due to what it described as the model’s exceptional ability to identify and exploit security vulnerabilities in software. It was not available to the general public.
Fable 5, released just three days before the ban, was Anthropic’s commercial answer to that power. The company equipped Mythos with strong guardrails, especially around cybersecurity and biology, and released it as a safer, publicly accessible version. According to benchmark testing from Vals AI (a firm that tracks AI performance), Fable 5 immediately became the most capable AI model available to the public at the time of its release.
In other words, the US government has just forced the shutdown of the world’s most advanced publicly available AI system.
The Jailbreak Claim, And Why Anthropic Disagrees
The government’s stated reason for the directive centers on a claimed “jailbreak” of Fable 5 , a technique that allegedly allows users to bypass the model’s safety guardrails.
Anthropic pushed back, firmly.
In a lengthy public statement, the company said the government provided only verbal evidence of what it described as a “potential narrow, non-universal jailbreak.” The technique in question essentially involves asking the model to read a specific codebase and identify software vulnerabilities, something Anthropic says is routine behavior used daily by cybersecurity professionals who protect systems, not attack them.
Critically, Anthropic also noted that the same capability exists in OpenAI’s GPT-5.5 and other publicly available models, without any similar government intervention.
Here’s what makes this even more pointed: Anthropic had spent weeks before Fable 5’s launch working with the US government, the UK’s AI Safety Institute (AISI), and multiple private red-team organizations, collectively logging thousands of hours testing the model’s safeguards. Those tests showed Fable 5 had “substantially more effective” guardrails than any previously deployed model.
“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)
The Bigger Political Context
This shutdown doesn’t exist in isolation. It’s the latest flashpoint in a rapidly escalating conflict between Anthropic and the Trump administration.
Earlier this week, Defense Secretary Pete Hegseth publicly accused Anthropic of “arrogance and betrayal,” after Anthropic CEO Dario Amodei refused to allow the Pentagon unrestricted access to Claude’s models for military use. Amodei drew a firm line, declaring the company “cannot in good conscience” permit its technology to be used for fully autonomous weapons systems or mass surveillance of American citizens.
The Trump administration responded by instructing all US government agencies to immediately stop using Anthropic’s technology, effectively blacklisting the company from federal contracts.
Then came Friday’s export control directive, which Anthropic complied with, while making it unmistakably clear it believes the government got this one wrong.
Anthropic has since filed a lawsuit against the federal government, alleging retaliation for its refusal to allow Claude to be deployed in lethal autonomous warfare.
What Does This Mean for AI Users Globally?
For everyday users, access to Claude’s other models (including Claude Sonnet, Claude Haiku, etc.) is unaffected. Only Fable 5 and Mythos 5 are offline. Anthropic has said it is working to restore access as soon as possible and believes the situation is a misunderstanding.
For Pakistani and international users specifically: This is a direct example of how US export control policy can affect AI access globally, overnight and without warning. Users, developers, and businesses that had integrated Fable 5 into their workflows found it simply gone on Friday evening, with no advance notice.
For the AI industry broadly, Anthropic itself warned that if the government’s standard, that any non-universal jailbreak justifies pulling a model offline, were applied consistently across the industry, it would “essentially halt all new model deployments for all frontier model providers.” That’s a significant statement about where AI governance is heading.
Anthropic’s Defense-in-Depth Strategy
One of the most technically interesting aspects of this story is Anthropic’s defense of its approach to AI safety.
Rather than claiming their model is un-jailbreakable (which they openly say is impossible for any AI today), Anthropic built what they call a “defense in depth” strategy:
- Make jailbreaks narrow rather than universal
- Make universal jailbreaks very expensive to produce
- Require 30-day data retention on Fable 5 deployments so the company can detect and shut down any successful attacks quickly
This is a mature, honest approach to AI safety, acknowledging that no perfect safeguard exists while building multiple layers of protection. Whether the US government accepts that argument remains to be seen.
What Happens Next?
Anthropic has promised to share more technical details about the claimed jailbreak within 24 hours of the shutdown and is actively working with government officials to restore access. The company’s lawsuit against the federal government adds a legal dimension that could take this battle into the courts.
For the AI world watching closely, the next few days will be defining. Either the government will accept Anthropic’s explanation and restore access, or this becomes a landmark case that reshapes how AI models are regulated, deployed, and controlled, not just in the US, but across the world.
Key Takeaways
- Anthropic has suspended Fable 5 and Mythos 5 globally following a US government export control directive
- The directive was issued on national security grounds, citing a claimed jailbreak of Fable 5
- Anthropic complied but publicly disagreed, calling the action disproportionate
- The shutdown is the first-ever federal intervention forcing a live commercial AI model offline
- All other Claude models remain fully operational
- The move is tied to a broader conflict over the Pentagon’s demand for unrestricted military use of Claude
- Anthropic has filed a lawsuit against the US government, alleging retaliation
Stay updated on the latest AI news, tools, and opportunities at BeingGuru.com, your go-to resource for AI, freelancing, and digital skills in Pakistan and beyo
