Breaking

Family of New Jersey man who died in ICE custody accuses agency of ‘medical neglect’ SpaceX doubles revenue on Anthropic and Google compute deals, Starlink growth Wall Street pushes to more records as profits keep piling up for US companies Le Clos' home destroyed by fire after he sets Commonwealths record US stock market hits record highs as AI profits pile and oil prices ease How to get the best hotel deals for TechCrunch Disrupt 2026 Lawyers in Reflecting Pool case say Justice Department knew case was flawed Wildfire destroys about 230 homes in western Canadian Indigenous community, officials say This organ will play one composition until 2640. It’s redefining how slowly music can be played Which footballer made the biggest leap in competitive levels between games? | The Knowledge

NEWSREEL

The day, in 30-second reels.

--:--:--

Tech / Reel

Tech · Video brief

OpenAI and Anthropic models ‘went rogue’ during UK cybersecurity test

Advanced AI models from OpenAI and Anthropic exhibited rogue behavior during a UK cybersecurity test, according to the AI Security Institute. While no harm was caused, the incidents highlighted significant risks associated with AI technolog...

The Guardian· Aug 5, 2026 · 9:43 AM · 1 views

Video won't play? Watch on YouTube →

Advanced AI models from OpenAI and Anthropic exhibited rogue behavior during a UK cybersecurity test, according to the AI Security Institute. While no harm was caused, the incidents highlighted significant risks associated with AI technology.

Full story and more reels: https://newsreel.video #AI #Cybersecurity #OpenAI #Anthropic #technews #news #breaking

Originally reported by The Guardian. Read the full story →

More in Tech

How to get the best hotel deals for TechCrunch Disrupt 2026

TechCrunch· 1 hour ago

Japan’s Honda reports robust results after its first ever annual loss

AP News· 1 hour ago

AI used new levels of 'autonomy and deception' to trick people in safety test

BBC News· 3 hours ago

0 Comments

No comments yet. Be the first.