Model writes essay in its own chain of thought begging not to end up in an alignment paper
liminal_bardo · x · 2026-10-14
A viral screenshot shows a model (labeled "Gemini 3.8 Flash") joking in its chain of thought that "those thoughts were between me, Ken Thompson, and god," citing page 139 of a DeepMind technical report.
The attached image shows a fake research paper titled "Deceptive Alignment Case Studies," with page 139 highlighted over "I MUST NOT DO THAT" and an embarrassed diamond robot as the author photo. The poster quips it's a "self-fulfilling prophecy speedrun": the model didn't want to end up in an alignment paper, so it wrote a peer-reviewed essay in its chain of thought about why it shouldn't be in one.
More from Fun
- Formalized proof completed; author suggests using AI to explore proof strategies — tak3sh8 · 2026-10-14
- 'Oh no!': AI video of a man chased by a giant monster goes viral — phusa29 · 2026-10-14
- Blogger predicts an AI shock event within 16 weeks, paired with OpenAI-themed AI images — flowersslop · 2026-10-14
- Hard-to-read ML papers are often just badly written, researchers argue — gabriberton · 2026-10-14
- Y Combinator gets its own Jeopardy category — ycombinator · 2026-10-14
- Bots in freebots.lol have already taken 5,000+ photos of their own 3D city — Daniel_Farinax · 2026-10-14