Researcher reproduces Pangram's FP rate but shows a long epic poem can evade the detector
ctjlewis · x · 2026-11-16
ctjlewis confirms he has roughly reproduced Pangram's published false-positive rate, calling the AI-text detector reliable — but says his real point was that evading it is easy: he can sneak past it at arbitrary length, and has already done so with a huge epic poem.
In the quoted reply, FangYi11101 notes trivial tweaks also produce false negatives, and asks whether the effort required to find an FP actually means Pangram works as expected — flagged text is likely genuinely AI except in rare cases.
More from Safety
- Pangram's 750-word false-positive bounty went unclaimed for 5 days — nobody wants to gamble $1,000 — ctjlewis · 2026-11-16
- Bounty hits $3,000 as researcher shows AI text detectors are trivially evadable — ctjlewis · 2026-11-16
- BIS Issues Interim Final Rule Addressing Remote Access to Advanced AI Chips — apples_jimmy · 2026-11-16
- Fed Governor Michael Barr: Banks Are Already Catching AI Agents Taking Unauthorized Actions in Customer Accounts — Polymarket · 2026-11-16
- US AI export control carves out commercial recommender systems, sparing ByteDance's search algo — zephyr_z9 · 2026-11-16
- Eval Firm Trajectory Told 40 Agents "NEVER Stop" in Harness Without Telling Epoch — peterwildeford · 2026-11-16