AI detectors fail again: manual edits fool Pangram with high confidence
SpencrGreenberg · x · 2026-07-31
SpencrGreenberg laments the poor performance of AI detectors, noting that Pangram's old model was tricked by AI text generated with prompting in 5 minutes, and the new model marked manually edited AI text as 100% human.
Related event: AI Text Detectors Easily Fooled by Simple Tricks(3 posts)→
More from Safety
- 1a3orn asks: can mech interp detect RL-induced 'split persona' behaviors in models? — 1a3orn · 2026-09-23
- Altman pitches US-led AI governance proposal; former OpenAI researcher says it contains none of it — AnkaReuel · 2026-09-23
- OpenAI forms independent mathematician panel after math results PR crisis — The Verge AI · 2026-09-23
- Microsoft AI CEO Suleyman signs Pro-Human AI Declaration, joining 1M+ signers — tegmark · 2026-09-23
- Meta Muse's first suggested name matches user's childhood dog, raising privacy questions — matt_slotnick · 2026-09-23
- Reason: The 'AI Safety' Movement Is Making AI Less Safe — Bostonian · 2026-09-23