US Government Orders Anthropic to Pull Fable 5 and Mythos 5 Models Offline
The unprecedented export control on Anthropic's advanced AI models marks a new regulatory phase, even as major labs secure record funding and valuations shift.
The story
In a significant and unprecedented move, the U.S. government on June 12 ordered AI lab Anthropic to take its recently released Fable 5 and more advanced Mythos 5 models offline. The directive, issued by the Commerce Department and signed by Secretary Howard Lutnick, cited national security authorities, requiring Anthropic to suspend access for all foreign nationals, both within and outside the U.S..
Anthropic, which had just released Fable 5 widely and maintained tight controls over Mythos 5 due to cybersecurity concerns, stated it received the order at 5:21 p.m. ET on June 12 and moved to comply, effectively disabling the models worldwide.
The company expressed disagreement with the handling of the matter, arguing the action lacked transparency and specific national security concerns, and that the identified 'jailbreak' capability is widely available in other models. This regulatory intervention comes as the AI sector sees immense capital flow, with Anthropic itself recently surpassing OpenAI in private valuation at $965 billion. The incident sets a new precedent for government oversight of advanced AI systems.
Who moved
Anthropic
What Changed: The U.S. government ordered Anthropic to take its Fable 5 and Mythos 5 AI models offline for all foreign nationals, leading to a global shutdown of these models.
Consequence: This marks the first instance of a leading AI lab being compelled to pull publicly deployed models due to government export controls, setting a significant regulatory precedent.
U.S. Government (Commerce Department)
What Changed: Issued an export control directive on June 12, forcing Anthropic to suspend access to its Fable 5 and Mythos 5 models for foreign nationals.
Consequence: The action represents the most significant step to date by the U.S. government to restrict access to advanced AI models, citing national security concerns.
OpenAI
What Changed: Made its frontier models and Codex coding agent available to Oracle Cloud customers through existing Oracle Universal Credits.
Consequence: This move expands enterprise access to OpenAI's technology, allowing Oracle Cloud Infrastructure customers to route AI workloads without separate procurement channels.
Prometheus (Jeff Bezos's industrial AI startup)
What Changed: Closed a $12 billion Series B funding round, valuing the company at $41 billion.
Consequence: The substantial capital infusion will fuel the development of an 'artificial general engineer' aimed at automating heavy engineering work and drug design in the physical world.
Tata Consultancy Services (TCS)
What Changed: Formed a global premier partnership with Anthropic and is training 50,000 associates on Claude.
Consequence: This partnership aims to accelerate the deployment of AI solutions into full production for highly regulated sectors like financial services and healthcare.
New models
Fable 5
Lab: Anthropic
What: A general-purpose model built on Anthropic's Mythos-class systems, designed for broader deployment with safeguards for high-risk requests.
Use: Aimed at complex, multi-step tasks with long-context reasoning, though currently offline due to government directive.
Mythos 5
Lab: Anthropic
What: An advanced autonomous model, more capable than Fable 5, initially with tightly limited access due to cybersecurity fears.
Use: Intended for sophisticated applications requiring high autonomy, but currently unavailable globally due to U.S. export controls.
Gemini 3.5 Pro
Lab: Google
What: A new iteration of the Gemini model.
Use: Planned for public rollout in June 2026, positioning it as a serious contender for multimodal and long-context work.
Market signals
Anthropic's private valuation surpassed OpenAI's, reaching $965 billion after a $65 billion Series H round.
Implication: This signals a dynamic shift in market leadership and intense competition among frontier AI labs, with valuation milestones rapidly changing.
The AI sector attracted $297 billion in funding during Q1 2026, accounting for 81% of all global venture activity.
Implication: This demonstrates a massive and concentrated influx of capital into AI, significantly eclipsing investments in other technology sectors.
Alphabet announced plans to raise $80 billion through public stock sales, including a $30 billion public offering and a $10 billion private placement with Berkshire Hathaway.
Implication: Major tech incumbents are actively securing substantial capital, likely to fund their extensive AI development and infrastructure initiatives.
Databricks reported over $5.4 billion in annual recurring revenue (ARR) with 55-65% year-over-year growth and positive free cash flow.
Implication: This provides a strong signal of profitability and scale within the AI software sector, setting a benchmark for future AI SaaS IPOs.
Builders saying
Dario Amodei, Anthropic CEO. He called for a more robust policy framework governing advanced AI, including government authority to block or reverse deployments that fail rigorous safety standards, a statement made particularly relevant by recent government action against Anthropic.
What we'll be watching
- Google's planned public rollout of Gemini 3.5 Pro in June 2026.
- OpenAI DevDay 2026, scheduled for September 29 in San Francisco.
- Anthropic's potential IPO, with discussions ongoing for a listing as early as October 2026.
- Databricks' anticipated IPO, expected in H2 2026 or early 2027.
- SpaceX's IPO, following a roadshow that began on June 8 and targeting a $1.75 trillion debut.
Reporting + analyst voices: grounded via Google Search at publish time.