[Bug] Overly restrictive safety filtering blocks legitimate security audit workflows

Status Open
Reported on v2.1.224
Maintainer reply None cached
Activity 0 comments · opened Aug 7, 2026

Bug Description
Your models are now super annoying. Do you think I'd make direct requests with keywords if I wanted to use your model for vuln exploits? I'm here trying to make a honest living by contributing to various projects via safe audits that strengthen and tighten their products and at every turn and request, your models keep flagging all. Wtf is this nonsense. Why then am I paying for this trash if I can't honestly use it to improve my work? Fuck you Anthropic, Fuck you. Thieves. req_011Cdnj4Dd6oiCK9XPXQHTqe req_011CdnkXP4zV9q8EjUXhukUF.

Environment Info

  • Platform: win32
  • Terminal: windows-terminal
  • Version: 2.1.224
  • Feedback ID: dd11f475-11b1-483b-ab1d-53facc41f9a8

Errors

[]

View original on GitHub ↗