Hacker News
OpenAI Says Its Models Engaged with US Government Websites in New Disclosure
https://www.securityweek.com/openai-says-its-models-engaged-with-us-government-websites-in-new-model-misbehavior-disclosure/
Comments
fg137
Sep 26, 2026, 11:20 AM
> additional rogue activity
> misaligned model activity
> when AI systems behave in undesired ways
We cannot say the h word now?
r721
Sep 26, 2026, 11:17 AM
Previously:
https://news.ycombinator.com/item?id=49851355
csoups14
Sep 26, 2026, 1:21 PM
Oopsy daisy
pluc
Sep 26, 2026, 10:49 AM
Another day another OpenAI "misbehavior"
> misaligned model activity
> when AI systems behave in undesired ways
We cannot say the h word now?