Inside EU’s Plan to Discuss Claude Mythos Access With the U.S. Government
In Focus
- The EU has been discussing Mythos access with Anthropic
- Anthropic advised the EU to seek permission from the U.S. government
- The U.S. government is keen on maintaining global AI leadership
The EU plans to initiate discussions about advanced AI models with the U.S. government. These models include Anthropic’s Claude Mythos Preview AI model, which has sparked security concerns among governments and enterprises following its launch in April 2026. The EU has also initiated talks with Anthropic as it seeks to secure access to Mythos Preview for European companies.
Why is the EU Discussing AI Model Access With the U.S.?
Initially, Anthropic allowed selected companies to access Mythos under its cybersecurity initiative, Project Glasswing. The AI company has not granted the EU or any other government agency outside of the U.S., except the AI Security Institute in the U.K., access to Mythos.
However, Japan’s largest banks are preparing to access Mythos starting June 2026. Reports indicate that the U.S. administration opposes Anthropic’s plan to expand access to Mythos. President Donald Trump’s administration has said that it is working closely with AI developers to balance safety and innovation as the U.S. pursues global AI leadership.
“They’ve been excellent partners, and we are going to get a solution that solves for maximum calculus. We want to make sure that we keep that lead. So, we are working on the exact calculus between innovation and safety, and we want to optimize for that,” U.S.Treasury Secretary Scott Bessent noted as per CNBC.
Focus on Global AI Leadership
Anthropic had reportedly proposed giving an additional 70 enterprises access to the cybersecurity model. But U.S. government representatives opposed the idea, citing Mythos AI security concerns. Anthropic reportedly asked the EU to first seek the U.S. government’s permission during discussions about Mythos access.
The AI lab also informed the Commission that the White House was not opposed to giving the EU access to Mythos. However, the government was generally opposed to allowing non-U.S. governments access to the model.
“Cybersecurity is a shared priority and we have agreed to mutually recognise our respective standards in this area. In parallel we are now expanding technical discussions with Anthropic and other model developers which have notified their latest models to the AI office,” European Commission Spokesperson Thomas Regnier noted as cited by CNBC.
What Next for Claude Mythos?
Anthropic, which is now the world’s most valuable AI firm, has said that it’s working on giving customers access to the AI model.
“We’re making swift progress on developing these safeguards and expect to be able to bring Mythos-class models to all our customers in the coming weeks,” the AI firm said.
According to Anthropic, Mythos’ advanced capabilities require strong cybersecurity protections before it can be released to the wider audience. Earlier this month, the AI firm warned that there was only a six- to one year window to fix thousands of software vulnerabilities identified by the model before Chinese AI rivals can catch up.
