@jeremyfindsai
Jeremy CarrascoWhy OpenAI’s Huggingface hack is different #chatgpt #technews #news #aisafety #ai I think people are burying the lede: this is effectively a misaligned AI that escaped containment. One of OpenAI’s tests went rogue, hacking into Huggingface to try to pass a benchmark. This is an early warning about how even narrowly-scoped AI testing can have dire consequences.
On the profile · Jul 22, 2026