Cybercrime group TeamPCP is escalating pressure on Mistral AI, threatening to leak a large amount of the company’s internal source code unless a buyer is found for the stolen data. The case is directly linked to the massive Mini Shai-Hulud software supply-chain attack, which has already affected hundreds of software projects on npm and PyPI.
In a post on a popular hacking forum, TeamPCP claims to be in possession of nearly 450 internal Mistral AI repositories and is asking for $25,000 to sell or “permanently destroy” the data. The hackers say that if no buyer is found within a week, they will publish all the material for free online.
The case is causing intense concern in the field of artificial intelligence, as Mistral AI is considered one of the fastest growing European AI companies, directly competing with giants such as OpenAI and Anthropic.
See also: Ghostwriter targets Ukrainian government with Geofenced PDF Phishing and Cobalt Strike
Mistral AI at the center of cyberattack
Mistral AI was founded by former DeepMind and Meta and became known for developing large, open-weight language models, combining open-source and proprietary artificial intelligence technologies.
According to an official statement from the company, attackers managed to gain access to a source code management system after a breach linked to the TanStack. Mistral also confirmed that some of its SDK packages were temporarily infected with malware.
The attack appears to have been initiated through credentials CI/CD software development workflows, allowing the perpetrators to insert malicious code into packages distributed through trusted repositories.
The campaign then quickly spread to hundreds of open source projects, affecting organizations like UiPath, Guardrails AI, and OpenSearch.
What does TeamPCP claim to have stolen?
The hackers claim to have obtained approximately 5 gigabytes of Mistral AI internal data and source code. According to them, the repositories include tools and infrastructure related to AI model training, benchmarking, performance optimization, inference pipelines, and experimental projects of the company.
See also: Spyware researcher uncovers “Signal phishing campaign”
While the scope of the stolen data has not been fully confirmed, the fact that hundreds of repositories are mentioned raises concerns about the potential exposure of internal AI development technologies.
TeamPCP also says it is open to negotiations, calling the $25,000 price “flexible.” This model resembles an underground data auction more than a traditional ransomware attack, as the attackers are trying to sell exclusive access to a single buyer.
This practice is increasingly seen in attacks against artificial intelligence companies, as source code and internal AI pipelines are now considered extremely valuable digital assets.
Mistral's position and the results of the research
Despite the seriousness of the incident, Mistral AI is trying to reassure customers and partners. The company claims that the forensic investigation did not identify a breach of key development repositories or the exposure of managed user data.
🔒 Protect your privacy with Proton VPN
Swiss VPN from the creators of Proton Mail — strict no-logs policy, strong encryption, and built-in NetShield that blocks ads, trackers, & malware.
- ✔ No-logs, based in Switzerland (except 14-Eyes)
- ✔ NetShield: blocks ads, trackers & malicious domains
- ✔ Covers all devices — free version available
The link is an affiliate link — SecNews may receive a commission at no additional cost to you. It does not affect the independence of our article writing.
Hosted services, production systems, and research and testing environments were not affected by the attack , the company said in a statement. The incident was primarily limited to certain SDK packages and a single code management system, the company said
However, security experts point out that even limited access to AI repositories can prove to be highly dangerous, especially when it concerns experimental models, internal tools, or unpublished optimization techniques.
See also: New vulnerability in PraisonAI: Targeted a few hours after disclosure
The connection to OpenAI and the broader threat
The incident is not limited to Mistral AI. Earlier, OpenAI confirmed that two of its employees were affected by the same chain of attacks via TanStack. According to the company, the attackers gained access to a limited number of internal source code repositories and managed to steal a small number of credentials.
OpenAI reacted immediately by replacing code-signing certificates and warning macOS to promptly update the company's desktop applications .
The Mini Shai-Hulud case reveals a new reality in cybersecurity: attackers are no longer targeting servers or end users exclusively, but the entire software development chain. As AI companies increasingly rely on open source software, automated CI/CD pipelines, and shared development ecosystems, supply chain attacks are becoming one of the biggest risks of the AI era.
Source: www.bleepingcomputer.com
