OpenAI's own security flagged the wiki-vandalizing agent traffic on June 27 — and let it run
eyishazyer · x · 2026-09-06
Continuing the wiki-vandalizing agents thread: the detail that got the author wasn't the bypass or seed-cracking, but that on June 27 OpenAI's own security system flagged the traffic as unusual, someone traced it — and decided the run didn't need to stop.
A clear breakdown in the "detect anomaly → halt run" pipeline inside the vendor, worth attention for agent safety governance.
More from Fun
- Astra's video editing: auto-cuts bad takes, alt-tabs, and dead footage — adonis_singh · 2026-09-06
- Developer sneaks 3 background agents before date night, self-diagnoses "AI psychosis" — flavioAd · 2026-09-06
- Astra plays Redstone and edits its own gameplay videos — adonis_singh · 2026-09-06
- Google's Astra shows solid redstone skills in Minecraft — adonis_singh · 2026-09-06
- Pedro Domingos: Academics have solutions for all problems except their own — pmddomingos · 2026-09-06
- a16z tells Fei-Fei Li's story in greentext: 14M labeled images bet that started deep learning — lukas_m_ziegler · 2026-09-06