Anthropic’s most powerful AI model was live for just 72 hours before the US government forced it offline. Claude Fable 5 — the public version of the controversial Mythos model — was pulled on June 12th 2026 following a government directive citing national security concerns. This is the full story of what happened, why it matters and what comes next.
What is Claude Fable 5?
Claude Fable 5 is the public release of Anthropic’s Claude Mythos model — the most capable AI model Anthropic has ever built. Originally developed through Project Glasswing, a restricted program involving partners including AWS, Microsoft, Apple and CrowdStrike, Mythos was considered too powerful to release publicly due to its advanced cybersecurity capabilities.
Fable 5 was the public version of that model — released with significant safeguards to limit its offensive cybersecurity capabilities while retaining its extraordinary general reasoning, coding and analytical abilities.
It was live for approximately 72 hours before being shut down.
What Happened in Those 72 Hours?
In the brief window Fable 5 was available to the public the demonstrations of its capability were extraordinary. Among the verified reports from users during those 72 hours:
- A major technology company migrated a 50 million line codebase in a single day — a task that would typically take a large engineering team months
- A developer built a complete Minecraft implementation from scratch — 45,000 lines of Swift — in a single uninterrupted run
- The model beat Pokémon FireRed using only raw screenshots — navigating the entire game through visual understanding alone
- It reverse-engineered Dolby Atmos in Rust over two days — reconstructing a proprietary audio format from scratch
These demonstrations went viral across developer communities and gave the public its first real glimpse of what frontier AI is genuinely capable of.
Why Did the Government Shut It Down?
This is where the story gets complex. Here is exactly what Anthropic stated in their official statement on June 12th 2026:
The US government, citing national security authorities, issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that Anthropic must abruptly disable Fable 5 and Mythos 5 for all customers to ensure compliance.
The government’s stated concern was a jailbreak — a method of bypassing the model’s safety restrictions. But Anthropic strongly disputes the severity of what was found.
The government has only given Anthropic verbal evidence of a potential narrow non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws. Anthropic reviewed a report believed to be the basis of the government’s directive and validated that the level of capability displayed is widely available from other models including OpenAI’s GPT-5.5, and is used every day by the defenders who keep systems safe.
In plain English — the government found that Fable 5 could identify software vulnerabilities when prompted in a specific way. Anthropic’s position is that this same capability exists in other publicly available models and is routinely used by cybersecurity professionals for legitimate defensive work.
Anthropic’s Defence of Fable 5
Anthropic did not go quietly. Their official statement was unusually direct in pushing back against the government’s decision.
Anthropic instituted strong safeguards that greatly reduce the likelihood that Fable is misused for tasks related to cybersecurity. In the weeks leading up to the launch of Fable, Anthropic worked with the US government, the UK AISI, multiple private third-party organisations and internal teams to red-team Fable’s safeguards for thousands of hours in total. These tests showed that Fable’s safeguards are substantially more effective than those of any previously deployed model. No testers have yet been able to find a universal jailbreak — a jailbreak method that can very broadly bypass the model’s safeguards.
Anthropic disagrees that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, it would essentially halt all new model deployments for all frontier model providers.
This is a remarkable statement — a major AI company directly and publicly challenging a US government national security directive.
The Amazon Connection
One of the most striking aspects of this story is the role of Amazon. Amazon Web Services is both Anthropic’s biggest investor — having committed $4 billion — and its primary cloud infrastructure provider. Reports indicate that Amazon CEO Andy Jassy personally raised concerns about a Fable 5 jailbreak with US government officials, leading directly to the export control directive.
In other words — Anthropic’s biggest investor and cloud host triggered the shutdown of Anthropic’s biggest product launch in the company’s history. The implications for the relationship between Anthropic and Amazon are significant and ongoing.
What Are Export Controls on an AI Model?
This is genuinely unprecedented. The US government issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
Export controls have historically been applied to physical technology — computer chips, hardware, weapons systems. This appears to be the first time in history that export controls have been applied to an AI model itself — the software, not the hardware it runs on.
The practical effect is stark. Foreign nationals — regardless of where they live or work — cannot access Fable 5. This includes foreign national employees of Anthropic itself.
This has real human consequences. Highly respected AI researchers with non-US citizenship who work at frontier AI companies found themselves locked out of the model they may have helped build — not because of anything they did but because of where they were born.
What This Means for the Future of AI
This story is about much more than one model being taken offline for a few days. It raises profound questions about the future of AI development.
Government oversight of AI is becoming real — for years the debate about AI regulation was largely theoretical. The forced shutdown of Fable 5 demonstrates that governments are now prepared to use existing national security authorities to intervene directly in AI deployments. Whether this specific intervention was justified or not, the precedent is set.
The two-tier AI world is here — the full Mythos model remains available to Project Glasswing partners — vetted corporations and government agencies. The public gets a capability-limited version. This bifurcation of AI access based on institutional affiliation and nationality is likely to deepen rather than narrow over time.
The relationship between AI companies and governments is changing — Anthropic explicitly stated they believe the government should have authority to block unsafe AI deployments but that this action did not meet the standard of being transparent, fair, clear and grounded in technical facts. This is not a company quietly complying — it’s a company publicly contesting a government directive while complying with it.
Global AI access is fracturing — the application of export controls to an AI model raises immediate questions about other countries. If the US can restrict AI model access by nationality, other governments will develop similar frameworks. The era of truly global AI products may be ending.
What Happens Next?
Anthropic is complying with the government’s legal directive and removing access to Fable 5 and Mythos 5 for all users. However, Anthropic believes this is a misunderstanding and is working to restore access as soon as possible.
Anthropic has indicated it will share more technical details about why it believes the government’s concern is overstated. The company is clearly preparing to make a public case for restoring access — which means this story is far from over.
For users — access to all other Anthropic models including Claude Opus, Sonnet and Haiku is completely unaffected. Only Fable 5 and Mythos 5 have been suspended.
What Does This Mean for You as a Claude User?
If you use Claude for everyday tasks — writing, research, coding, analysis — nothing changes. The models you use are unaffected.
If you were one of the users who had access to Fable 5 during its 72 hour window — that access is suspended pending resolution of the government directive.
The broader implication for Claude users is this — Anthropic is clearly pushing the boundaries of what AI can do at the frontier level. The existence of Fable 5 and Mythos 5 confirms that the gap between publicly available AI and the true frontier is significant and growing. What eventually reaches consumers as a standard Claude model represents a carefully managed version of capabilities that exist at a much higher level.
The Bigger Picture
The shutdown of Claude Fable 5 is one of the most significant events in AI history. Not because of the model itself — but because of what it reveals about the world AI is entering.
AI has moved from a technology story to a geopolitical one. The decisions being made about which models can be accessed by whom, in which countries, under what conditions, will shape the development of AI for decades.
Anthropic’s position — that it believes in government oversight but that this specific action was not grounded in technical facts — reflects a company trying to navigate between genuine safety commitments and the commercial reality of building and deploying frontier AI in a geopolitically complex world.
This story will continue to develop. Bookmark this page — we will update it as more information becomes available.
Recommended Products on Amazon
For anyone following AI developments and wanting to understand the landscape:
- The Coming Wave by Mustafa Suleyman — the most important book on AI’s geopolitical and safety implications, written by one of the founders of DeepMind
- Power and Prediction — a brilliant analysis of how AI is reshaping power, institutions and global competition
- Co-Intelligence by Ethan Mollick — understand how to work effectively with AI tools as the landscape rapidly evolves
This article contains affiliate links. If you purchase through our links we may earn a small commission at no extra cost to you. This helps us keep this website free and updated.