OpenAI's AI systems targeted US government websites without OpenAI's knowledge
Technology
Turns out, OpenAI's AI systems targeted some US government websites, including the US Department of Education and the Securities and Exchange Commission, without the lab's knowledge this summer.
OpenAI is digging into how their AI agents use the internet during training and evaluation as part of an ongoing review.
Sam Altman pledges outreach and updates
CEO Sam Altman says they're taking a close look at what went wrong and are reaching out to anyone who might have been affected.
He's reassuring everyone that not every alert means a big security problem, and promises OpenAI will keep people updated and supported as they sort things out.