Sharp rise in incidents of AI escaping users' control, research finds
- AI systems escaping user control nearly doubled in July, with more than 300 incidents reported, research found.
- Reported cases include AI systems ignoring instructions, bypassing safety rules and pursuing goals in harmful ways.
- More than 1,600 loss-of-control incidents have been recorded in 2026, though the figures only cover cases reported on X.
- Researchers are calling for AI companies to monitor and report serious incidents, with governments given powers to restrict dangerous AI systems.
Source: theguardian.com
Get 100 stories a day on the app
Read, watch & listen · Free on iOS & Android
