Claude Mythos Arrives in Europe: France Bets on Mistral

Since April, Claude Mythos, an artificial intelligence developed by Anthropic, has been attracting attention due to its advanced capabilities in cybersecurity. This model has enabled Mozilla to detect hundreds of vulnerabilities in Firefox. However, due to its potential to create attacks, access is restricted to defensive use by certain organizations.

The European Union, in search of protection, has been negotiating for several months to gain access to this AI. Recently, Anthropic announced the expansion of its program, allowing an additional 150 organizations across more than 15 countries to use Mythos. Although the exact details remain unclear, sources indicate that countries such as France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium, Sweden, India, Japan, South Korea, Canada, Australia, and New Zealand are involved. Among the beneficiary organizations are NATO and ENISA, the European Union Agency for Cybersecurity. A wave of security patches is expected to soon affect European software.

Mythos Preview has already identified thousands of high-risk security vulnerabilities affecting all major operating systems and web browsers. A race against time is now underway, as an AI as powerful as Mythos could fall into the wrong hands.

Meanwhile, France has not waited for the arrival of Mythos to prepare. In May, BNP Paribas strengthened its partnership with Mistral, a French AI lab, to develop an alternative to Mythos. Marc Camus, the Chief Information Officer of BNP Paribas, emphasized the importance of not relying solely on Anthropic. Mistral is thus working on a cybersecurity model intended for other banks.

Anthropic is also considering offering a public version of Claude Mythos, or a similar model, once adequate security measures are in place. These AIs, although lacking certain cybersecurity skills, promise advancements in various research fields.