Europe Edges Closer to Claude Mythos Access

Artificial Intelligence & Machine Learning , Geo-Specific , Next-Generation Technologies & Secure Development Europe Edges Closer to Claude Mythos Access Anthropic Offers ENISA a Place in Project Glasswing David Meyer • June 1, 2026     Share Post Share Credit Eligible Get Permission Image: Shutterstock Anthropic offered the European Union’s cybersecurity agency ENISA entry to Project Glasswing, its arrangement for giving organizations controlled early access to its vulnerability-finding Mythos artificial intelligence model. See Also: How AI Increases the Risk of Enterprise Data Exposure ENISA spokesperson Laura Heuvinck confirmed the offer on Monday. Bloomberg reported earlier in the day that Anthropic would give the agency access to Mythos. "They have invited us to have access and we are looking at the way the potential access would work - the conditions and so on," Heuvinck told ISMG. It is now nearly two months since Anthropic revealed the existence of Mythos Preview, sparking a flurry of activity in the cybersecurity sector. Mythos marked a step change in the ability of generative AI models to spot and exploit software vulnerabilities. There is widespread concern that, once such a model sees wider release, it will give unprecedented hacking powers to regular people. Anthropic for months did not give any EU-based organizations access to Glasswing and Mythos. This caused intense pressure on the European Commission from lawmakers who said the continent was being left vulnerable. Commission Executive Vice President Henna Virkkunen said in May that public and private organizations would just have to make do with "already-available advanced cyber tools" to scan their systems for bugs (see: Europe Prepares to Hunker Down Against Bug Finding AI Models). In mid-May, Anthropic rival OpenAI said it would give the commission access to its new GPT-5.5-Cyber model, which boasts similar capabilities to those of Mythos, per analysis by the U.K. government's AI Security Institute. OpenAI is also restricting access to its advanced bug hunting model. It is unclear at this time which European agency can use the new OpenAI model. The commission’s fledgling AI Office will gain the power to demand starting in August, but ENISA is also a possibility. Heuvinck said Monday that ENISA is in discussions with Anthropic and OpenAI over finalizing the terms of access. She said the agency and the AI companies each have conditions, but did not specify what those were. "We hope to have some more news in the coming days or weeks," Heuvinck said. Anthropic did not responded to a request for comment. The advent of the new models has caused particular concern in the financial sector, with the European Central Bank, the International Monetary Fund and the German financial regulator Bafin all issuing urgent warnings in the last couple weeks (see: ECB: AI Means European Banks Must Hasten Cybersecurity Pace). European banks still don’t have access to Mythos, but France’s Mistral has developed a rival to that model and is talking to banks such as Paris-based BNP Paribas about giving them tools to harden their defenses, German media outlet Heise reported. "The game changer is the speed at which we have to address vulnerabilities and the scale. There are lots of them discovered at once," BNP chief information officer Marc Camus said at a press conference with Mistral last week. The two companies announced a deepening partnership on cybersecurity. "So we need to prepare ourselves for that and that's something we are really working on ⁠very, very hard," Camus said.