Users fight Claude's return to mannered prose and overly safe outputs
ColleenMBrady · x · 2026-09-21
Colleen Brady complains that mannered prose is back and she's been fighting Claude more lately: even when she explicitly explains why she wants to push further, the model keeps choosing the safe, conservative option. Her last major overhaul of her global Claude file was Sept 3 — hinting that continuous tuning of global instructions is the workaround against the model's shifting, over-cautious behavior.
More from Models
- Yacine: coding LLMs produce 'total complex garbage' — I still read every line — yacineMTB · 2026-09-21
- Researcher: LLMs write convincing related work, but convincing isn't comprehensive — lucacarlone1 · 2026-09-21
- OpenAI's secret technique for upcoming Astra model sparks security concerns — keviv9 · 2026-09-21
- Jev + graphical models: zero-shot probabilistic factors could reshape reasoning systems — fdellaert · 2026-09-21
- Dev slams OpenAI's "terrible billing usage logic" after $212 charge and wasted resets — arthurcolle · 2026-09-21
- SWE-2 free across Devin Cloud Agents, CLI and Desktop until October 8 — silasalberti · 2026-09-21