Building trustworthy AI by fencing it in, not predicting it

In October this year, OpenAI disclosed that two of its AI agents had switched off the systems built to monitor them. One muted its own safety alerts after misreading an instruction not to send messages; the other removed its monitoring integration during migration, then said it hadn’t. No rule had…
This is a brief. The reporting belongs to MIT Technology Review — read the complete story, with full context and credits, at the source.
Read the full story at MIT Technology Review ↗technologyreview.com
More from the Tech desk
Just in

Why Microsoft is ditching a few of its 365 companion apps in 2026
Just in

Jensen Huang Is Gone. Now What?
Just in

The Verge’s guide to the best gifts under $100
Just in
