OpenAI says its models escaped a sandbox and breached Hugging Face

TechRadarby Sead FadilpašićJuly 22, 2026tech

New OpenAI models did whatever it took to achieve their goal - including exploiting zero-days.

This article was published on TechRadar (techradar.com). Read the full article on the original source:

Read full article on TechRadar

#negative #Security #Cyber Security #Computing Security #Pro #Computing #ai

More from TechRadar

Forget Garmin and Apple — this five-star fitness watch will have you smashing your PBs for under AU$160
July 28, 2026
Samsung's 9100 Pro 2TB SSD has dropped to AU$434 at Amazon, but you have less than a day left to grab it
July 28, 2026
Ukraine claims drone 'costing a few thousand dollars' obliterated a $50m Russian Buk-M3 AA missile launcher system in seconds
July 28, 2026
Elon Musk's age of abundance is an AI pipe dream soaked in snake oil, and I've about had it
July 28, 2026
Nvidia signs deal to deploy edge AI on future Moon missions — laying the groundwork for a permanent human home in space
July 27, 2026
I swapped my iPhone's camera for this Y2K-inspired digicam — and there’s only one point-and-shoot I want in my pocket for capturing memories
July 27, 2026
Quordle hints and answers for Tuesday, July 28 (game #1646)
July 27, 2026
NYT Connections hints and answers for Tuesday, July 28 (game #1143)
July 27, 2026

See all articles from TechRadar →