[Bug] False positive safety flag on Russian-language planning text with automatic model downgrade
Open 💬 0 comments Opened Jul 6, 2026 by jplaksina29-pixel
Bug Description
This appears to be a false positive on a Russian-language message. The message was ordinary planning text, not coding, cybersecurity, biology, or unsafe content.
Flagged text:
“Какой берем — А (с фазой B на старте), Б или В?”
The model was switched automatically from Fable 5 to Opus 4.8. Please review and tune this safeguard for Russian-language conversations.
Environment Info
- Platform: darwin
- Terminal: cursor
- Version: 2.1.201
- Feedback ID: 013259e7-82f5-476c-884a-95f39c456a3a
Errors
[]