Thinking Machines exec's tweet flagged by Pangram AI detector — he posts Gdoc history to prove it's human
luke_drago_ · x · 2026-11-11
A delightfully meta AI-detection false positive: austinc3301 spotted that a tweet by Luke Drago got a low-confidence Pangram AI flag — while wearing the Pangram badge itself.
- Drago posted his Google Docs version history showing 2-3 hand-written drafts of the flagged sentence, cut down for character count
- He confirms no AI was used, joking that he "could have written it better if it's getting flagged"
- The flagged tweet was promo copy for Thinking Machines hiring policy researchers; heavy human editing (deleted clauses, restructured sentences) appears to be the detector's blind spot
More from Companies & People
- Vals AI hackathon adds PokerBench track with $7,500 first-place prize — rhythmrg · 2026-11-11
- Boring Company started tunneling in San Antonio just 29 days after finishing Bastrop — McDonaghMatthew · 2026-11-11
- Grok Build Left 35,000+ User Apps Exposed, Including Medical Records and SpaceX Sales Plans — Promptmethus · 2026-11-11
- Model PMs Earn 60% Premium Over Peers; Under 75 People Globally Have the Depth — aakashgupta · 2026-11-11
- Anthropic's ex-Claude Code product lead: 10 great evals beat hundreds — aakashgupta · 2026-11-11
- YC F26 startup Petrichor launches a wet lab that runs parallel experiments for frontier AI labs — ycombinator · 2026-11-11