GitHub Hacked: 3,800 Repos Stolen via Malicious VS Code Extension — OpenAI and Mistral AI Also Hit
🔗 The attack chain — how it all started
This affair is a perfect example of a cascading supply chain attack : a single initial point of compromise contaminated dozens of organisations within days, like a digital domino effect.
Here is the precise sequence as confirmed by GitHub and security researchers:
- Step 1 — TanStack compromised: Hackers infected the TanStack/router npm package, a very popular JavaScript library, by exploiting weaknesses in the publishing process
- Step 2 — Nx Console infected: One of the Nx team developers had their workstation compromised via TanStack. Hackers then published a malicious version of the Nx Console VS Code extension
- Step 3 — Millions of devs infected: Developers worldwide updated Nx Console without suspecting anything — silently downloading a GitHub token-stealing malware
- Step 4 — Access to private repos: With the stolen tokens, hackers were able to access the private GitHub repositories of their victims — including employees of GitHub itself, OpenAI, Mistral AI and Grafana Labs
🏢 Who was affected?
😱 Why this is particularly serious
What makes this attack exceptionally dangerous is what is inside the internal GitHub repositories of companies like OpenAI or GitHub itself:
- Proprietary source code — the internal algorithms of OpenAI and Mistral AI models
- API keys and secrets — credentials giving access to entire cloud infrastructures
- Training data — the datasets used to train AI models
- Internal vulnerabilities — unpatched flaws in their products
- Infrastructure blueprints — the technical architecture of the world's most secure systems
- Customer data — depending on how the repositories were organised
💡 What this means for you — even if you're not a developer
This attack doesn't directly affect everyday users — but its consequences concern you indirectly and deeply.
- If you use ChatGPT (OpenAI) — internal code may have been stolen, which could eventually enable fake versions of ChatGPT to be created to trick you
- If you use Le Chat (Mistral AI) — second incident in a week for Mistral, raising questions about the platform's overall security
- If your company uses Grafana — infrastructure monitoring data may have been exposed
- For everyone — this attack shows that even the world's most secure companies can be compromised via a simple development tool
- Change your GitHub password immediately
- Revoke all your personal access tokens on GitHub: Settings → Developer settings → Personal access tokens → Delete all
- Check your VS Code extensions — uninstall Nx Console if you have it, and any extension you didn't intentionally install
- Audit your repositories for unauthorised access: Settings → Security → Audit log
- Enable two-factor authentication on GitHub if not already done
🌍 The 2026 wake-up call — supply chain, the new Achilles' heel
This affair is part of a deep trend in 2026: supply chain attacks have become the preferred attack method of the most sophisticated hackers. Rather than directly attacking ultra-secure targets like GitHub or OpenAI, they infect tools that these targets use.
In 2026, supply chain attacks have increased by 312% compared to 2024. TanStack → Nx Console → GitHub/OpenAI/Mistral is just the latest example. The same technique hit SolarWinds in 2020, XZ Utils in 2024, and now the entire JavaScript/npm ecosystem in 2026.
❓ Questions fréquentes
Questions about the security of your digital tools?
CyberGuard answers all your questions — free, no jargon, in plain English.
🤖 Talk to CyberGuard →