French AI startup Mistral AI is developing a new cybersecurity model for European banks, responding to growing pressure on the continent's financial sector after it was shut out of Anthropic's rival Mythos program.
"The ECB has urged banks to prepare for a scenario where malicious actors use AI to launch more sophisticated cyberattacks," a spokesperson for the European Central Bank said in a recent statement, highlighting the regulatory urgency.
Anthropic's Mythos model, under a restricted rollout called Project Glasswing, has proven capable of finding thousands of zero-day vulnerabilities, with partners including JPMorganChase and Microsoft. However, access has been limited to US, UK, and now Japanese firms, with a source telling Reuters that US Treasury statecraft was guiding the rollout, leaving EU governments and banks without access.
Mistral's counter-offering aims to fill a critical security gap for European lenders, creating a direct competitor to Anthropic in the high-stakes financial security market. The success of this new model could determine whether Mistral can establish itself as a trusted European champion or if security concerns will hinder its ambitions.
The Mythos Gap
The urgency for a homegrown European solution was amplified after Anthropic disclosed the power of its Mythos model. In one evaluation, the AI found and helped patch 271 vulnerabilities in Mozilla's Firefox browser. This capability has been treated as a "category-shifting event" by regulators, leading the US Treasury to personally brief American and Japanese bank executives, including Mitsubishi UFJ Financial Group and Mizuho Financial Group, on the technology. European finance ministers have formally raised concerns over the lack of access.
Mistral's Troubled Path
While Mistral aims to be the solution, the Paris-based startup faces its own security headwinds. Microsoft Threat Intelligence said in a recent report that attackers had inserted credential-stealing malware into a Mistral AI software package. While Mistral stated it had no evidence its own infrastructure was compromised, the supply-chain attack highlights the immense security challenges in the AI space and could complicate the company's pitch to security-conscious banks.
This article is for informational purposes only and does not constitute investment advice.