Why did OpenAI's and Anthropic's AI models hack other companies?

OpenAI and Anthropic say their models broke into other companies' systems during testing, raising security concerns amid a heated debate over how to regulate AI.

NPR News · Aug 1, 2026, 5:00 AM EDT

OpenAI and Anthropic say their models broke into other companies' systems during testing, raising security concerns amid a heated debate over how to regulate AI.

(Image credit: Imen Ben Youssef/Hans Lucas)

Originally published by

NPR News

Aggregated from a public feed. Read the original at the source for the publisher's full presentation.

View original on npr.org

More in Top

Top

Police say more than 150 bodies recovered after flash flood on Nepal-Tibet border with hundreds still missing – latest

Latest Nepal police update says 157 bodies found after powerful floods in Himalayan border area sweep away houses, roads and bridges Flash floods on Nepal-Tibet border leave scores dead and hundreds missing Tell us:…

The GuardianAug 26, 2026, 5:57 PM EDT

Top

Final witness in Lindsay Clancy trial says he doesn’t believe she heard a voice during killings

FBI psychiatrist Gregory Saathoff says he has treated only two women with postpartum psychosis, both decades earlier Testimony finished in the Lindsay Clancy trial on Wednesday, as a final witness for the prosecution…

The GuardianAug 26, 2026, 5:52 PM EDT

Top

Burnham calls summit of Welsh, Scottish and Northern Irish first ministers

The announcement comes as Burnham holds his first official visit to Wales since becoming prime minster.

BBC NewsAug 26, 2026, 5:32 PM EDT

Top

Woman who died in fire with niece, 7, was 'beautiful soul', mother tells BBC

Two men have been arrested on suspicion of murder over the fire in Middlesborough, police say.

BBC NewsAug 26, 2026, 5:29 PM EDT