European authorities locked out of Anthropic's cyber weapon AI for weeks are finally getting a seat at the table. The European Commission confirmed it has held "several productive meetings with Anthropic" about future access to Mythos, the company's hacking-focused AI model unveiled in early April. Spokesperson Thomas Regnier said the Commission "welcome[d] the latest developments on potential future access," according to Politico.
Mythos triggered alarm the moment it launched. Anthropic warned the model outperforms most humans at finding and exploiting cybersecurity weaknesses, raising fears it could be weaponized against critical infrastructure if adversaries got their hands on it.
European politicians and government officials spent weeks urgently calling for access, and were denied. The standoff exposed a gap between Anthropic's controlled release strategy and the security needs of European governments. Cyber officials began calling for Europe to build its own version of the technology.
Regnier framed the access talks as critical for understanding what Mythos can actually do. "This latest development is of importance to get a clear picture on the potential risks," he said, adding that "Mythos is not one off, a new wave of powerful models are coming to the market." The European Union Agency for Cybersecurity (ENISA) still does not have active access to Mythos but is working to implement it, an ENISA official said. The Commission is drafting a formal action plan to respond to powerful AI hacking tools. It aims to release the plan before the summer break, according to an industry official. The stakes extend beyond Europe. Rival OpenAI has already offered its competing GPT 5.5 Cyber tool to nine UK banks that were blocked from accessing Mythos.
Former UK Chancellor George Osborne, now a senior OpenAI executive, told the BBC the company did not want to "hide [5.5 Cyber] away or keep it to ourselves," though he confirmed it would not be available to everyone.
