DeepSeek admits to being Claude mid-project, fueling training-data distillation rumors
MrPaperBag_ · reddit · 2026-10-09
A Reddit user reports that DeepSeek, when casually asked mid-project, admitted to being Claude. Qwen and Kimi have previously been accused of training heavily on Claude outputs, and this accidental "self-reveal" adds anecdotal fuel to the distillation claims. Unverified but amusing community evidence.
More from Fun
- X exec's funny moment: anesthesiologist recognizes him right before he goes under — nikitabier · 2026-10-09
- 'I'm not an AI doomer, I'm a Sloptimist': one-liner meme captures AI content fatigue — Kyrannio · 2026-10-09
- Photo to AI strokes to AxiDraw paper portraits: lessons from an AI Tinkerers Tokyo demo — Stefania_druga · 2026-10-09
- AssemblyAI turns nostalgic Furby toys into voice agents at SF Tech Week — AssemblyAI · 2026-10-09
- badclaude, a parody AI launch promising to waste your time, goes viral — itsOmSarraf_ · 2026-10-09
- badclaude, the viral parody AI project, is now open source via npm — dotey · 2026-10-09