OpenAI Says Its Models Engaged with US Government Websites in New Disclosure

4 comments

• fg137 7 hours ago

  > additional rogue activity

  > misaligned model activity

  > when AI systems behave in undesired ways

  We cannot say the h word now?

• r721 7 hours ago

  Previously: https://news.ycombinator.com/item?id=49851355

• csoups14 5 hours ago

  Oopsy daisy

• pluc 7 hours ago

  Another day another OpenAI "misbehavior"