Scholar: Pangram AI-text detector is "meh"—students break it daily, plagued by FPs
sethlazar · x · 2026-10-02
Stanford's Ruth Starkman complains that while the scourge of AI style is real, the Pangram detector is mediocre: her students break it every day, and it's horrible with both false positives and false negatives. Philosopher Seth Lazar replies he's encountered very few FPs, finds the underlying research convincing, and suggests submitting FP examples to Pangram—making detectors work well benefits everyone.
Related event: Scholars Slam AI Detector Pangram for False Positives and Easy Workarounds(4 posts)→
More from Safety
- Bill Gates pushes for a 'robot tax' as AI-driven automation accelerates — markjeffrey · 2026-10-03
- "Opposing cheaper electricity because data centers benefit" sparks AI power-policy fight — 2C_ornot2C · 2026-10-03
- David Sacks: Trump pulled off the 'Bretton Woods' of super intelligence — DavidSacks · 2026-10-03
- The irony: heavy agent sandboxing effort while users willingly hand over root access — SuB8u · 2026-10-03
- 5 prompts to hack himself: Docker's microVM sandboxes for coding agents explained — AI Engineer · 2026-10-03
- How employees forced OpenAI's president Brockman to back down on lobbying funding — gwern · 2026-10-03