Key Takeaways
- Commerce Department ended its two-week block on Claude Mythos 5, allowing release to 100+ US institutions
- The decision establishes a new federal oversight framework for frontier AI model releases
- OpenAI released GPT-5.6 to government-approved partners on the same day
The US government has lifted its block on Anthropic's Claude Mythos 5, allowing the company to release its most powerful AI model to more than 100 American institutions. Commerce Secretary Howard Lutnick signed the authorization Friday, ending a two-week standoff that shut down both Mythos and its consumer-facing counterpart, Fable 5.
The timing is pointed. OpenAI released its GPT-5.6 to a similar list of government-approved partners the same day.
Why did the US block Mythos in the first place?
Two weeks ago, the Trump Administration imposed export controls on Mythos after Amazon and other companies warned that the model could be jailbroken for malicious purposes. Both Mythos and Fable 5 went dark. Fable had briefly been the most powerful AI model available to everyday consumers.
The deeper concern: Semafor reported that US officials believed Anthropic had released Mythos to partners too closely linked to China, specifically referencing a South Korean telecom provider. That connection triggered the national security response.
What does the new arrangement require?
Lutnick's letter to Anthropic's chief compute officer Tom Brown cited "significant progress" from daily negotiations. Under the new terms, organizations listed in the letter's Annex A no longer need export licenses to access Mythos 5. That includes the organizations' foreign national employees and Anthropic's own foreign staff.
Anthropic committed to work with the government on "protocols and standards and releases" for its models going forward. Translation: the company agreed to let Washington weigh in on how and when it ships frontier AI.
Commerce spokesman Benno Kass framed the speed as a feature: "In just two weeks, we have worked diligently to ensure America remains the global leader in AI while safeguarding our security."
What happens to Fable 5?
Lutnick's letter says nothing about Fable 5. People close to the talks told Semafor that negotiations are moving toward a Fable release, but no timeline exists. For now, consumers and smaller businesses remain locked out.
A new federal regime for AI releases
This decision does more than resolve one company's problem. It creates precedent. The Commerce Department is now building, on the fly, a framework that gives federal officials control over when and to whom frontier AI models ship. Lutnick's letter reads like a license, not a one-time exception.
Lab leaders have pushed back. Losing even days in the global AI race carries real costs when competitors in China and elsewhere are sprinting. But the government's position is clear: national security trumps speed-to-market.
European officials and other US allies are frustrated. They depend on decisions made in Washington with little visibility into the process. Non-US companies and governments still don't know when they'll get Mythos or Fable access.
Who gets access to Mythos 5 now?
The letter grants access to more than 100 US institutions, including major companies and government agencies. The full Annex A list hasn't been published. But the framing, "trusted partners," suggests a curated set of defense contractors, cloud providers, and strategically important enterprises.
This isn't open access. It's controlled distribution to organizations the government believes can secure the model against misuse.
Logicity's Take
This move sets the template for how frontier AI will be regulated in the US. Expect every major lab, OpenAI included, to face similar checkpoint negotiations before releasing their most capable models. For enterprise buyers, the lesson is practical: if you want early access to the next generation of AI capabilities, you need to be on the government's trusted list. That likely means cleared facilities, compliance infrastructure, and documented security controls. Smaller companies and startups should watch how Fable 5 plays out. If consumer-tier models face the same friction, the gap between enterprise AI and everyone else widens.
The competitive pressure
OpenAI's same-day release of GPT-5.6 to approved partners isn't coincidence. Both labs are racing, and both are now operating under government supervision. The question is whether this framework helps or hurts American AI companies against foreign competitors who face no such constraints.
Anthropic has raised over $6 billion from Amazon and Google. OpenAI has Microsoft's backing. Neither company will run out of capital. But if every model release requires weeks of negotiation with Commerce, the development cadence slows. That's the tradeoff Lutnick's letter codifies.
Frequently Asked Questions
What is Anthropic's Claude Mythos 5?
Mythos 5 is Anthropic's most powerful AI model, positioned above its consumer-facing Fable 5. The US government restricted its release over national security concerns before granting access to approved institutions.
Why did the US government block Mythos 5?
Officials cited jailbreaking risks and concerns that Anthropic had released the model to partners linked to China, reportedly including a South Korean telecom provider.
Which organizations can access Mythos 5 now?
More than 100 US institutions listed in the Commerce Department's Annex A, including major companies and government agencies. The full list hasn't been made public.
When will Fable 5 be available again?
No timeline has been announced. Negotiations are reportedly ongoing, but Lutnick's letter did not address Fable 5.
Does this affect OpenAI and other AI labs?
OpenAI released GPT-5.6 under a similar government-approved framework the same day, suggesting all frontier labs now operate under federal oversight for major releases.
How AI tooling companies are building proprietary capabilities amid the model wars
Need Help Implementing This?
If you're evaluating frontier AI models for enterprise deployment, understanding compliance and security requirements is now essential. Contact Logicity for analysis on how federal AI oversight may affect your technology roadmap.
Source: Hacker News: Best
Non-US nationals cleared for Mythos 5 under new exception
The new article provides a specific detail about the exception: non-US nationals who are Anthropic employees or members of approved organizations are cleared to use Mythos 5. It also notes that Anthropic is actively working with the government to expand Mythos 5 access and restore Fable 5, though no timeline is given, while OpenAI expects 'a few weeks' for GPT-5.6 Sol.
What Claude Mythos can actually do: Inside Anthropic's cybersecurity AI and Project Glasswing
The new article reveals Mythos's specific technical capabilities: it achieved a perfect score on Cybench, discovered thousands of unknown software flaws including a 17-year-old FreeBSD remote code execution vulnerability, and was used in 'Project Glasswing' - a collaborative effort with tech giants to secure critical software. It also provides the specific release date of April 7th, 2026 and explains Anthropic's reasoning for restricting access was specifically about offensive cybersecurity capabilities, not general misuse concerns.
Huma Shazia
Senior AI & Tech Writer
Produced with AI assistance and reviewed by the Logicity editorial team. Learn more in our Editorial Policy.
