Anthropic will offer the EU use of its cutting-edge Mythos AI model to plug cybersecurity holes.
Mythos was unveiled but not released in April after it was found to be able to detect — and exploit — security flaws in almost all systems.
Anthropic provided it to a small group of crucial, mainly US-based organizations to help find and fix vulnerabilities. But no EU-based bodies were included, leading to concerns that the bloc would be left exposed in the new, AI-driven cybersecurity era.
The problem is urgent: Although Mythos itself remains mostly under wraps, similarly powerful products are coming, including open-source Chinese models, which are less than 12 months behind the frontier and will soon bring comparable capabilities to anyone, including hacking groups.
