IdeaLens: a new detector that tells whether a text's ideas came from a human or an AI
MohitIyyer · x · 2026-10-06
Existing AI detectors assume a text's author also came up with its ideas — increasingly untrue, as humans may brainstorm while AI writes (or vice versa).
Mohit Iyyer's team launched IdeaLens, a demo that checks both the ideas and the words in a piece of writing, responding to emerging AI-use policies that focus on who originated the ideas rather than who wrote the text.
The demo is live while credits last, and the team especially wants users to report failure cases directly on the site.
More from Safety
- Researcher warns of wave of mass automated AI-powered hacking attacks in coming months — kevinnbass · 2026-10-06
- The handoff test: approve, revoke, transfer to a fresh agent—does human authority survive? — tallmetommy · 2026-10-06
- Swiss poll: 94% want binding international AI rules as a diplomatic priority — S_OhEigeartaigh · 2026-10-06
- Australia, Britain & Norway move to ban smart glasses in public over privacy — Polymarket · 2026-10-06
- Letting AI act on your behalf shouldn't shield you from liability — dhadfieldmenell · 2026-10-06
- China moves to curb AI companions over emotional dependency concerns — Olivier__OG · 2026-10-06