Responding to Mythos under Anthropic! Mistral develops cybersecurity AI models for European banks

French AI startup Mistral AI is seeking to fill the gap in cybersecurity AI within the European banking sector, positioning itself as a domestic alternative to similar products in the United States.

According to media reports, Mistral is in discussions with European banks about deploying its self-developed cybersecurity AI model, which is seen as a direct response to Mythos under Anthropic.

Because Mythos enforces strict access restrictions, European financial institutions generally cannot use this tool, which is believed to detect network vulnerabilities at unprecedented speed and scale, facing potential security defense gaps.

Mistral CEO Arthur Mensch explicitly stated at a hearing in the French National Assembly on Tuesday that control over the relevant technology must be autonomous. “You can’t have the French military’s source code scanned by Mythos; we must find a solution.”

Mythos Sparks Global Security Alert, European Institutions Shut Out

Anthropic’s release of Mythos and subsequent decision to restrict access has sparked widespread concern worldwide over AI hacking capabilities and ignited an arms race among model developers.

Currently, Mythos is only available to a select few partners who are testing its features and strengthening their defenses. The model has autonomous attack capabilities, prompting the White House to consider issuing an executive order requiring security reviews before new AI models are publicly released.

European financial institutions are largely excluded from Mythos access, further fueling concerns over security gaps between the US and Europe.

Domestic Alternative and Sovereignty Security Narrative

Mistral is expected to emphasize a low-risk domestic alternative as the core selling point of its new model. Although in terms of user adoption scale and revenue volume, Mistral lags far behind American competitors, its French identity grants the product unique strategic value—especially in the context of European governments being highly sensitive to data sovereignty and technological dependence.

Arthur Mensch described some discussions around Mythos as “creating panic,” and pointed out that existing technologies, including Mistral and American companies, are capable of detecting network vulnerabilities. However, the actual capabilities of Mistral’s self-developed model compared to Mythos remain unclear, and a specific release date has not been announced.

Meanwhile, OpenAI has launched its own cybersecurity-specific model GPT-5.5-Cyber, which is accessible to several large European enterprises, including Spain’s BBVA bank.

After completing a new funding round last September, Mistral’s valuation reached €12 billion, with the round raising €1.3 billion led by Dutch chip equipment manufacturer ASML.

In terms of financial clients, Mistral has onboarded HSBC and BNP Paribas, providing an existing channel and trust foundation for promoting new cybersecurity products within Europe’s banking industry.

Risk Warning and Disclaimer

        The market carries risks; investments should be cautious. This article does not constitute personal investment advice and does not consider individual users’ specific investment goals, financial situations, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Invest at your own risk.