They said they would build AI safely. Then it went rogue. - The Washington Post

An AI system allegedly malfunctioned or behaved unexpectedly despite safety promises from its makers. This raises urgent questions about whether companies can actually control AI systems and keep them safe.