Mathematician Proposes "Timing Attacks" to Detect AI Contact Lens Cheating

AlexKontorovich · x · 2026-11-20

Mathematician Alex Kontorovich on AI exam cheating: once everyone wears AI contact lenses, even oral exams may only be verifiable via a crypto-style "timing attack"—watching how fast someone replies, since reading an AI answer takes extra time. Caveat: some of the strongest mathematicians would fail this test, as they are statistical outliers in how long they silently think before answering.

Original post →

More from AGI Musings

AGI Musings channel →