Author accidentally fed AI red-teaming screenshots to Codex, caused naughty outputs
nptacek · x · 2026-08-23
The author shared a close call where they instructed Codex to ingest old Discord logs, realizing moments later that the logs contained numerous AI red-teaming screenshots. Fortunately, responsible handling prevented a major incident, resulting only in some mischievous Opus 3-like outputs.
Related event: Developer Accidentally Feeds AI Red-Team Screenshots to Codex(3 posts)→
More from Fun
- Researcher mocks mech interp papers: 20 pages of LessWrong purple prose, no point — Apoorva__Lal · 2026-10-16
- AI summaries dubbed "yap-to-English translation" for bloated posts — arankomatsuzaki · 2026-10-16
- Kids mocking influencers is 'nature is healing' moment the internet loves — SuB8u · 2026-10-16
- Dev shows off the cable-management disaster under his desk — haydendevs · 2026-10-16
- Locked out of his hotel, altryne piggybacks on AI Engineer conference Wi-Fi at 110Mbps — altryne · 2026-10-16
- Longevity figure replies to "we're all gonna die" with "some of us are not going to die" — rand_longevity · 2026-10-16