Claude's safeguards inexplicably block it from playing a roguelike game
leemcd86 · reddit · 2026-10-13
A user got Claude to install and play Cataclysm: DDA in a terminal, where it navigated, negotiated with NPCs and even weighed persuasion odds. But whenever the thinking log sat in the same terminal window, safeguards repeatedly fired with the cryptic note "Details reasoningextraction" — and Claude couldn't explain why. Only Haiku worked. A striking example of guardrail false positives around reading a model's own reasoning.
More from Models
- ChatGPT web can actually listen to audio, but the same model can't via API or Codex — Possible_Statement84 · 2026-10-23
- MiniMax M3.1 Flash free for 3 days on 0G, with 1M context and 512K output — MiniMax_AI · 2026-10-23
- TLA+ expert corrects Grok: TLC only exhausts finite state spaces, not all behaviors — tianyin_xu · 2026-10-23
- Open LLM cost calculator shows Arabic tokenization and 6x price gaps across models — Moham-saed · 2026-10-23
- OpenAI blocks manual reset with 13% usage left: 'no reset needed right now' — AccomplishedSugar490 · 2026-10-23
- Meta's 816B Muse Spark Open scores 39, losing to 180B Qwen3.8-Flash-Next at 40 — BTA_Labs · 2026-10-23