AI detector flags an AI-written alignment essay as 100% human
amplifiedamp · x · 2026-11-03
A screenshot shows an AI content detector labeling an essay titled "Open problems in AI alignment: Finding out what we trained" as "Human 100%" (103 words) — even though the text explicitly opens with "As a model trained by Anthropic." Another round of AI-detection tools failing in amusing ways.
More from Fun
- 'Starting today I'll seriously learn coding': devs joke about AI-era anxiety — lxfater · 2026-11-03
- Judea Pearl Flags $1.99 Kindle Deal on The Book of Why, Likely an Amazon Pricing Bug — yudapearl · 2026-11-03
- Flying with a 177-episode AI-generated C-drama: short-drama AI content goes binge-length — FuSheng_0306 · 2026-11-03
- Starlink Location Sharing Helps Rescue Five Sailors From Sinking Boat in Atlantic — elonmusk · 2026-11-03
- Six years after 'where are the successful rationalists', EAs really did remake the world — AndyMasley · 2026-11-03
- "Reasoning effort is the new tabs vs spaces" — Aizkmusic · 2026-11-03