On the evening of June 12, 2026, millions of users tried to access two of the world’s most powerful AI models, and got nothing. Anthropic announced it had disabled access to its Fable 5 and Mythos 5 models to comply with an export control directive from the US government that cited “national security authorities.”
” No prior warning. No explanation. Just a blank screen where a cutting-edge AI used to be.
This wasn’t a technical glitch. It was the first government-forced shutdown of deployed frontier AI models in history, and it changed everything about how we think about AI, power, and who controls the tools reshaping the world.
What Are Fable 5 and Mythos 5?
To understand what was switched off, you first need to understand what these models were, and why they mattered.
Fable 5 and Mythos 5 launched just three days earlier, on June 9, 2026, and Anthropic touted them as state-of-the-art across a number of different industry benchmarks. Their arrival was a landmark moment. Fable 5 marked Anthropic’s first general release of a model in the Mythos class, a tier above its Opus line, made possible by new safeguards that block responses in specific high-risk areas.
The two models weren’t identical. Built on the same underlying architecture, they differ primarily in their output controls: Fable 5 includes classifiers designed to block responses in high-risk areas such as cybersecurity, while Mythos 5, available to a separately vetted set of organisations, operates with some of those constraints removed. Think of Fable 5 as the powerful but guardrailed version available to everyone, and Mythos 5 as its less restricted sibling, accessible only to vetted partners in areas like cybersecurity research.
Both had been stress-tested before launch. Before release, Anthropic subjected the models to thousands of hours of red-teaming by the US government, the UK AI Security Institute, and third-party organisations, none of which found a universal jailbreak. The models passed. And then, three days after launch, Washington pulled the plug anyway.
What happened on June 12?
The sequence of events was startlingly fast. Anthropic received an export control directive from the US government, citing national security authorities, ordering the suspension of all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including Anthropic’s own non-citizen employees.
Commerce Secretary Howard Lutnick sent Anthropic CEO Dario Amodei a letter outlining the restrictions. The catch? The letter did not provide specific details of its national security concern.
Anthropic faced an impossible compliance problem in real time. There is no reliable way to segment foreign nationals from US persons in real time across a user base in the hundreds of millions, so the company turned the models off for everyone.
Why did the Government act?
The letter gave no specific national security details, Anthropic was left to piece together the reason on its own.
The company’s best understanding of events is that the government believed it had come across a method of bypassing, or “jailbreaking,” Fable 5. Specifically, a technique involving asking the model to read a specific codebase and fix any software flaws.
Anthropic pushed back hard. The company reviewed what it believes is the report underlying the government’s action and concluded that the level of capability displayed is widely available from other publicly deployed models, including OpenAI’s GPT-5.5, and is used every day by the defenders who keep systems safe.
There is no version of a capable coding model that can fix vulnerabilities but cannot also describe them.
According to multiple media reports, Amazon CEO Andy Jassy first raised concerns about the model with senior administration officials after Amazon researchers used a series of prompts to get the Mythos-class model to provide information about cyberattacks that were supposed to be restricted. Amazon, one of Anthropic’s largest investors, appears to have been the party that set the shutdown in motion.
Critics noted a further irony. The models built on the release of Claude Mythos Preview, which had captivated Wall Street and government officials with its advanced cybersecurity capabilities. By marketing Mythos as a very advanced cybersecurity tool, Anthropic may have handed the government the very legal grounds it needed to treat the model like a weapon.
The real-world fallout
For millions of ordinary users, the shutdown was baffling. For enterprises, it was a crisis. Enterprise clients in finance, healthcare, SaaS, and critical infrastructure found their core intelligence services abruptly disabled, without exception, prior warning, or effective recourse.
The disruption cascaded across the entire cloud ecosystem. Platforms spanning AWS Bedrock, Google Cloud, Microsoft Foundry, Snowflake, and the direct Claude APIs were all simultaneously affected. Businesses that had woven Fable 5 into daily workflows, from document processing to customer communications to code review, found those workflows simply gone.
The legal fine print offered little comfort. The event exposed gaps in enterprise contracts, as most SLA and force majeure clauses had never anticipated an instant government-mandated AI shutdown.
The timing could hardly have been worse for Anthropic’s own fortunes. The shutdown added disruption to Anthropic’s business at a delicate moment: the company had just filed a confidential IPO prospectus with the Securities and Exchange Commission, disclosing a revenue run rate of USD47 billion and a valuation of USD965 billion following a recent fundraising round.
What this signals for the future of AI
This was not the first time Anthropic found itself in Washington’s crosshairs. After negotiations between the two organizations collapsed, the Department of Defense declared Anthropic a “supply chain risk”, a label historically reserved for foreign adversaries, requiring defense contractors to certify that they would not use Anthropic’s Claude models in their work with the military. In an effort to reverse its blacklisting, litigation is still ongoing.
For a company carrying a nearly USD1 trillion valuation into a public offering, it has now been blacklisted twice by the federal government, forcing investors to consider whether that mega-valuation is fully considering the fact that the government is willing to switch off its flagship product overnight.
Anthropic has been consistent in its response. The company called the export directive a “misunderstanding” and stated that the process does not adhere to the principles of being “transparent, fair, clear, and grounded in technical facts.”
Historic, significant and taking shape
The move marks the first time the US government has used export controls to halt access to a commercial AI model already widely used by the public. That is a precedent. Once a government has used a power, it can use it again.
The geopolitical consequences may be equally significant. The unprecedented step has sparked concern from politicians around the world and intensified calls for sovereign AI, the idea that countries should control the AI models, infrastructure, and data that underpin critical technology, rather than depend on a small number of US AI labs.
Foreign customers may also drift toward homegrown options, China has its own frontier models, and Europe has no real answer. So, rather than protecting US AI dominance, this kind of intervention could accelerate the very fragmentation it fears.
Analysts also note a “chilling effect” on research – OpenAI might now think twice before shipping its next model, wary of losing billions in revenue to the same kind of order.
Some observers, however, quietly welcomed the moment. Some AI safety advocates saw the shutdown as a rare instance of meaningful AI deceleration, proof that the brakes exist, even if the car is already moving at speed.