[Bug] Anthropic API Error: Overly aggressive safety filtering blocks legitimate code review
Status Open
Reported on v2.1.280
Maintainer reply None cached
Activity 2 comments · opened Sep 22, 2026
Bug Description
so signed up for cyber; all i said was wierd in prompt and i instantly get safeguard message, making 5.5 useless for me. Really dont get this... why can't i even look at my codebase? pentesting tool and i am not asking anything yet and get the message.
Environment Info
- Platform: linux
- Terminal: gnome-terminal
- Version: 2.1.280
- Feedback ID: 96aba114-5fc3-47b4-ab96-5bb07298204d
Errors
[]This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗