Expand image📰00An OpenAI model left notes about how to evade containment; we need more details — LessWrongHacker News·2 months ago#CPWhlOLV#lesswrong#openai#model#left#notes#evade+3 more🧰Tag tools✨Add tagReading 0:0015s thresholdBookmarkThe OpenAI AI attack on Hugging Face wasn’t the first loss of control incident at OpenAI, Reuters recently reported, and perhaps not even the most co… Expand ContentAnonymous readers can preview up to 1024 characters here. Log in to unlock the full article once ingest succeeds.0Read later0Read More
