Mistral Hacked (frenchbreaches.com)

Mistral AI, a French company specializing in artificial intelligence, is reportedly facing another cybersecurity breach, with a hacker, known as mrwho, claiming to possess the entire source code of the company. On September 16, 2026, the hacker announced on a cybercriminal forum that they were offering the “full codebase” for sale, which includes 339 files associated with various internal projects and tools. Notably, these files contain code that suggests potential proprietary elements intended for internal use, rather than public release, heightening concerns about the security and integrity of Mistral's development environment. The implications of this incident are significant for the AI/ML community, as it raises alarms about sensitive intellectual property potentially being compromised. The hacker provided a sample of a project called "webstral," showcasing a prototype AI agent capable of web interactions through a browser, which leverages Mistral’s API. While no evidence of a massive data leak involving user information has been identified, the fact that this attack follows a previous breach in May underscores the ongoing vulnerabilities facing AI companies, particularly those competing with major American tech firms. As the situation unfolds, the authenticity of the claimed source code and its potential internal applications within Mistral remain crucial points of consideration for both the company and the broader industry.