📚 New to this topic? Read our full guide: What Is Artificial Intelligence?.
Anthropic Gets US Approval to Bring Back Its Claude Mythos 5 Model
The US government has said yes to Anthropic Claude Mythos 5 again. The company shared this news. Anthropic is an American AI company. AI (artificial intelligence) means computer software that can read, write, and think in a human-like way.
Mythos 5 is said to be Anthropic’s strongest AI tool for cybersecurity. A “model” is the trained program that does the thinking. Cybersecurity means keeping computers and networks safe from hackers and attacks.
This comes from a report by The Decoder. US groups that run and protect critical infrastructure can use the model again. Critical infrastructure means the key systems a country needs, like power grids, water, and banks. The approval ends a ban that had stopped people from using the model.
What exactly changed
The US government let Anthropic “redeploy” Mythos 5, the company said. Redeploy just means switch it back on for real use. This is allowed for American groups that run and defend critical infrastructure.
The Decoder says this permission likely matches one given to OpenAI. OpenAI is the rival company that makes ChatGPT. OpenAI was allowed to give its new top model, GPT-5.6 Sol, to a small set of groups and customers. A “frontier model” is one of the most advanced AI systems you can get.
There is also a people side to this. Under this rule, Anthropic workers who are not US citizens can use Mythos 5. The same is true for people at approved groups who are not US citizens. In short, the rules no longer stop non-American staff at these trusted groups from using the tool.
Why the model was restricted
Mythos 5 and another model, Fable 5, were both blocked earlier by government orders, The Decoder says. The report does not give the full reason this time. But it does make the timeline clear. Since June 12, Anthropic had been working with the US government to bring back both models.
This is a rules story, not a launch story. It is about an AI company getting the okay to turn a tool back on. It is not about new features or speeds. So we do not list benchmark scores or tech specs here, because the source does not give any. (A benchmark is a test score that shows how well a model performs.)
Key facts at a glance
| Detail | What the source reports |
|---|---|
| Company | Anthropic (US AI firm, maker of Claude) |
| Model approved | Claude Mythos 5 (its most powerful cybersecurity model) |
| Who can use it | US organizations running and protecting critical infrastructure |
| People cleared | Non-US nationals at Anthropic and at approved organizations |
| Work started | June 12, 2026 (talks with the US government) |
| Still pending | Wider access to Mythos 5; bringing back Fable 5; no timeline given |
| Comparable case | OpenAI cleared for GPT-5.6 Sol; expects “a few weeks” |
Who benefits
The first clear winners are US groups that defend critical infrastructure. They get back a top tool for cybersecurity work. Anthropic wins too, since a blocked model makes no money.
The Decoder adds that Anthropic is still working with the government on two more goals. The first is to let more people use Mythos 5, beyond this first group. The second is to make Fable 5 widely available again. The company has not given a date for either step. For comparison, OpenAI thinks its own wider rollout will take “a few weeks.”
Why it matters (especially for India and founders)
This story shows something big. Who can use the best AI models is now shaped by government rules, not just by what a company wants to sell. Top “frontier” tools can be switched off or limited by rules, and then switched back on under conditions.
For founders and businesses outside the US, including in India, the lesson is about planning. If your product depends on one powerful model, a rule change in another country could hurt you. The source does not name India or any country in the approved groups. So the simple takeaway is to be careful, not to feel safe. Build with backup options. Watch how “trusted partner” access grows over time.
It also shows where the AI race is going. Cybersecurity is becoming a front line. Both Anthropic and OpenAI are getting special approval for their strongest models in this exact area. That tells us governments see these tools as serious help for protecting key systems.
FAQ
What is Anthropic Claude Mythos 5?
It is an AI model from Anthropic, an American AI company. The Decoder calls it the firm’s most powerful model for cybersecurity. That means protecting computers and networks from attacks.
Who can use Mythos 5 now?
US groups that run and protect critical infrastructure. The approval also covers people who are not US citizens working at Anthropic and at approved groups.
What about Fable 5?
Fable 5 was also blocked earlier. Anthropic is working with the government to make it widely available again. But no date has been shared, says The Decoder.
Does this approval include India?
The source does not name India. It only talks about US groups and non-US citizens at approved groups. Any wider access beyond that is still being worked out.
The takeaway
Anthropic’s most powerful cybersecurity model is back online for a trusted set of US users. The ban is lifted. But full access for everyone is not here yet. Fable 5’s wider return and broader Mythos 5 access are still in progress, with no date set. The big picture: in 2026, who gets to use the strongest AI is more and more a government choice.
This story is closely tied to how governments and companies now guide AI use. See our related coverage on how ICAI is building an AI model with Sarvam AI for chartered accountants and on Microsoft extending free Windows 10 security updates to 2027.
Source: The Decoder (reporting by Matthias Bastian, June 27, 2026).
