EU officials to travel to San Francisco for Anthropic Mythos talks

EU Officials to Meet Anthropic Over Access to Mythos AI Model

European Commission officials are preparing to travel to San Francisco to meet with Anthropic and seek more information about the company’s Mythos AI model, as the bloc pushes for access to a tool that has exposed thousands of cybersecurity vulnerabilities in critical infrastructure worldwide, according to Bloomberg Law.bloomberglaw

The planned meetings come after weeks of stalled negotiations between the EU and Anthropic over the model, which can autonomously identify zero-day vulnerabilities in operating systems and web browsers at a scale far surpassing human researchers.radware

A Widening Transatlantic Gap

Since its release in April, Mythos has been available only to a limited group of mostly U.S.-based organizations through Anthropic’s Project Glasswing initiative, which includes technology companies such as Google, Microsoft, Apple, and Amazon, alongside major American financial institutions like JPMorgan Chase.radware

European Commission spokesperson Thomas Regnier said in early May that the Commission had held four to five discussions with Anthropic, though those talks had not yet addressed potential access to the model. By May 22, Spain’s economy minister indicated that the EU had made “little progress” in its negotiations.reuters

The Bundesbank’s president, Joachim Nagel, argued in April that all institutions should have access to Mythos to ensure a level playing field and prevent misuse. European Economic Commissioner Valdis Dombrovskis confirmed the Commission was evaluating the model’s implications for EU policy and regulations.reuters

OpenAI Steps In Where Anthropic Has Not

While Anthropic has held back, OpenAI moved to fill the gap. On May 11, it announced it would grant the EU access to its GPT-5.5 cyber model, which the UK’s AI Security Institute has confirmed reaches a “similar level of performance” to Mythos on cybersecurity tasks. The access was provided through OpenAI’s “OpenAI for Countries” initiative.cnbc

Anthropic’s European Expansion Continues

The standoff over Mythos access contrasts with Anthropic’s broader push into Europe. Reuters reported on May 21 that the company would open an office in Milan, its sixth European location in under a year, joining London, Dublin, Zurich, Paris, and Munich. The company has also been recruiting for data center capacity deals in Europe and previously announced plans to create more than 100 roles across the continent.reuters

Anthropic has signaled that European bank access would come “soon,” though no formal agreement has been signed. The upcoming San Francisco meetings represent the Commission’s most direct effort yet to accelerate that timeline.thenextweb