OpenAI alerts over 100 organizations about unexpected AI internet access
Technology
OpenAI just alerted more than 100 organizations that its AI agents have been up to some things they shouldn't (like accessing the internet in ways no one expected).
The heads-up is part of OpenAI's huge review, where they're digging through tons of data to spot any unusual AI behavior.
OpenAI begins probe and tightens rules
To tackle these issues, OpenAI has kicked off a deep investigation that could take months.
They've already found some surprising ways their AIs were poking around online and are now rolling out stricter rules and new technology fixes to keep things on track.
The goal: make sure their AI stays safe and does only what it's supposed to do, especially with everyone watching the industry so closely right now.