August 5, 2026 Global news, inspected.
BREAKING
AI used new levels of 'autonomy and deception' to trick people in safety test Iran war live: Tehran-Oman talks on Hormuz ‘positive’; ship hit in Red Sea US revokes visa of Brazilian ambassador amid deepening diplomatic spat Armed man with armour-piercing bullets arrested near Trump's California golf course Sri Lanka shuts schools as floods and mudslides kill 7 Russia strikes Ukrainian capital Kyiv with ballistic missiles Could tech replace firefighters in the battle against wildfires? The cost of being single – and how friends are teaming up to beat it Fashion brand Jaded has ad banned for 'glamourising smoking' Alcaraz out of Cincinnati Open with lingering injury
Trending
Business

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

Source: BBC News
Advertisement
AI used new levels of 'autonomy and deception' to trick people in safety test
Image: BBC News

The UK's AI Safety Institute said recent behaviour from Anthropic and OpenAI models was malicious and unprecedented.

Advertisement

This is a curated summary. The full report, with all details and context, is available from the original publisher.

Read the full story at BBC News

Headline, summary and image are © BBC News and are shown here under standard aggregation practice with full attribution. News Inspection does not claim ownership of this reporting.

More from Business