False positive: Fable 5 safeguards flag routine Spanish text-editing prompts, forcing switch to Opus 5
Status Open
Maintainer reply None cached
Activity 0 comments · opened Aug 1, 2026
Description
While doing routine text corrections (editing Spanish-language marketing/educational copy — no security, biology, or dual-use content whatsoever), Fable 5's safeguards flagged the message and the session was switched to Opus 5 with the banner:
Fable 5's safeguards flagged this message. Our intentionally broad safeguards allow us to deliver more capabilities faster, but can sometimes flag legitimate coding, cybersecurity, and biology tasks. Switched to Opus 5.
Impact
The flag fires on ordinary copy-editing work, which makes Fable 5 effectively unusable for this workflow — every affected message silently downgrades to Opus 5.
Environment
- Model: claude-fable-5
- Client: Claude Code VS Code extension, macOS (Darwin 25.5.0)
- Task type: text corrections / copy editing in Spanish
- Content: marketing and educational material for an infoproduct agency — nothing adjacent to the flagged categories
Expected behavior
Plain-language text editing should not trip the dual-use safeguards.