All incidents / Microsoft
Microsoft AI security incidents
2 security incidents involving Microsoft's AI since January 2025, including 0 where an agent acted on its own. The model was judged at fault in 2. Involvement does not mean Microsoft was responsible.
| 6 Oct 2026 | Adversa showed an encrypted web page could make Copilot CLI send out a .env file GitHub | Manipulated AI | Low |
| Jun 2025 | EchoLeak: zero-click prompt injection could make Microsoft 365 Copilot leak data Microsoft | Manipulated AI | Low |