Mistral AI in talks with European banks to rival Anthropic's Mythos cybersecurity model

Mistral AI, the French startup valued at €12 billion after raising €1.3 billion from investors including ASML, is in early talks with European banks about deploying a cybersecurity-focused AI model. The move is a direct response to Anthropic's Mythos, a powerful vulnerability-finding AI that is only available to a select few U.S. partners—leaving European institutions largely locked out. Mistral has been developing its own model and is working on an off-the-shelf version for wider rollout. The company already works with financial clients like HSBC and BNP Paribas on AI-driven security flaw detection.

Mistral CEO Arthur Mensch told a French National Assembly hearing that Europe must control this technology, warning against scenarios where “the French military’s source code scanned by Mythos” creates irreparable dependency. He dismissed the hype around Mythos as “fear-mongering,” noting that many other firms, including OpenAI (which released GPT-5.5-Cyber to European banks like BBVA), already have similar capabilities. While it's unclear how Mistral's model will compare to Mythos—which can perform autonomous attacks and prompted White House interest—Mistral is positioning itself as a lower-risk, homegrown alternative for Europe's banks.