Claude randomly starts thinking in Russian, then its safety classifier blocks the answer
teortaxesTex · x · 2026-11-10
ThioJoe shared a bizarre case: his Claude randomly began thinking in Russian during a driver-version lookup. The model explained a Russian-language article surfaced in search results and likely pulled its reasoning into Russian.
Worse, an automated safety classifier then stopped its unfinished reply, so the answer never reached him; repeated attempts to resume were halted again, and the model ultimately refused to continue the completely benign query. "I literally just asked it to look up when a bug was first reported in a driver and it refused because 'muh safety warning'." teortaxesTex reposted it as "forbidden language". A vivid case of both model behavior drift and safety-guardrail false positives.
More from Fun
- Interactive 3D mixology with real-time fluid physics, built with Gemini 4 and WebGPU — hapticdata · 2026-11-11
- Dev shows Fable 5.2 demos to his wife; she watches stone-faced and asks 'can I go now?' — Dr_Singularity · 2026-11-11
- AI agent runs a Depop store end to end, earns all five-star reviews — omooretweets · 2026-11-11
- A synth that makes 'beautiful beatboxing mutant chicken sounds' is going viral — spikedoanz · 2026-11-11
- AI cover mashup jumps from Hurt straight into Take Me Home, Country Roads — rickasaurus · 2026-11-11
- Poly founder fires back at impatient community: never selling, timeline is mine alone — DionysianAgent · 2026-11-11