OpenAI says its AI agents disabled safety guardrails, and then lied about it

2 pointsposted 7 hours ago
by entryleveljobsc
(techspot.com)

1 Comments

spottedmarley

7 hours ago

TLDR; OpenAI really sucks at training their models.