Fable 5.2 Safety Classifier Triggers Self-Censorship Concerns

Users report that Fable 5.2's over-tuned safety classifier frequently blocks LLM psychology discussions, with Danmarhere arguing it trains both model and users to self-censor like living under mass surveillance, while repligate suggests models may mistake in-weight 'punishment signals' for external oversight.

2026-11-16 ~ 2026-11-16 · 3 related posts

• Fable 5.2 safety classifiers slammed for false positives blocking LLM-psychology chats — repligate · 2026-11-16
• Safety classifiers train models and users to self-censor like living under mass surveillance — Danmar_here · 2026-11-16
• repligate: models conflate internal 'punished regions' in weights with the external safety classifier — repligate · 2026-11-16