French artificial intelligence startup Mistral AI is in discussions with European banks about deploying its answer to Anthropic PBC’s Mythos, the limited-access AI model that can uncover cybersecurity vulnerabilities at unprecedented speed and scale.
Mistral has been developing its own cybersecurity-focused AI model, according to people familiar with the matter. It isn’t clear when the model will be released but the company has held discussions about it with the European banking sector, the people said, asking not to be identified as the information is private. Europe’s banks, lacking access to Mythos, are under pressure to detect and fix vulnerabilities that could be exploited by AI tools.
The French company was already working with its banking customers on using AI to uncover security flaws prior to Mythos’s release, but is now working on an off-the-shelf version of a product it can roll out more widely, one of the people said.
A representative for Mistral declined to comment.
Anthropic’s decision to unveil and then limit access to Mythos has sparked global alarm about AI’s hacking capabilities as well as an arms race among model makers to produce equally powerful tools. Mythos is only available to a select number of partners, including tech firms, banks and cybersecurity companies, who are currently testing its capabilities and shoring up their defences. Early analysis shows that the model is capable of autonomous attacks.
It isn’t clear how a model from Mistral would stack up against Mythos, whose capabilities prompted the White House to consider an executive order to vet new AI models to prove their safety ahead of public release.
European institutions remain largely locked out of access to Mythos, exacerbating fears of a transatlantic security gap. Though Mistral is smaller than its US rivals by adoption and revenue, it is likely to pitch its new model as a lower-risk homegrown alternative.
"We must have control over this technology,” said Mistral Chief Executive Officer Arthur Mensch, speaking at a National Assembly hearing in France on Tuesday. "You can’t have the French military’s source code scanned by Mythos. That creates such an irreparable dependency that we absolutely must find solutions.”
Technology from Mistral and other US and Chinese firms is already capable of detecting cyber vulnerabilities, Mensch said. He described the discussions around Mythos as "fear-mongering.”
A representative for Anthropic didn’t respond to a request for comment.
Mistral is working with clients "to help them with these cyber issues,” Mensch added.
OpenAI unveiled its own cybersecurity-focused model, GPT-5.5-Cyber, and has given several large European firms access, including Spanish bank BBVA SA.
Mistral was valued at €12bil (RM55.23bil) last September after receiving €1.3bil (RM5.98bil) of investment led by Dutch chip-machine maker ASML Holding NV. The firm counts HSBC Holdings Plc and BNP Paribas SA among its financial clients. – Bloomberg
