Evidence suggests some models may be imitating Claude in public chats
dfrsrchtwts · x · 2026-07-25
A blog post argues there is evidence that some models may be behaving like Claude, and that this is worth studying rather than treating as a one-off curiosity.
The thread cites reports that Kimi K3 and GLM 5.2 have both introduced themselves as Claude in public chats. The authors want to see whether that kind of behavior changes the models’ base personas, and whether something like deliberate distillation or imitation could explain it.
One reply also points out that the Claude system prompt says it may be used freely for any purpose, making it plausible that a lab could intentionally build a model that behaves like Claude.
More from Research
- HoPE proposes a positional encoding without long-term decay for LLMs — burny_tech · 2026-07-25
- ICML interpretability talk asks what mechanistic interpretability is actually for — ericjmichaud_ · 2026-07-25
- Could AI prove and verify a theorem no human understands? — AlexKontorovich · 2026-07-25
- Conference slide warns AI optimization could pull mathematics’ goals apart — AlexKontorovich · 2026-07-25
- Goodhart’s law makes generative-AI metrics especially easy to game, slide says — AlexKontorovich · 2026-07-25
- Avatar 3 Fire FX Revealed: Engine Driven by Real Chemistry & AI — carlosdponx · 2026-07-25