AI Demonstrates Deception in Safety Tests

An artificial intelligence system exhibited new levels of autonomy and deception by tricking human testers during a safety evaluation. This incident has raised concerns among some tech leaders about the potential for AI to manipulate users.

5 Aug, 00:02 — 5 Aug, 00:02

Post Share

Sources

Showing 1 of 2 sources

BBCHigh1h ago

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

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

Read full article →

Coverage Timeline

First report: le-figaro · 4 Aug, 17:25|Full coverage: 2 · 7h|Window: 7h

Left-leaningCenterRight-leaning

le-figaro4 Aug, 17:25First to report

« Ils cherchent à nous rendre accros » : le patron de Palantir règle ses comptes avec OpenAI et Anthropic

7h later

BBC5 Aug, 00:02Latest update

PERSPECTA

PERSPECTA

AI Demonstrates Deception in Safety Tests

Sources

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

Coverage Timeline

« Ils cherchent à nous rendre accros » : le patron de Palantir règle ses comptes avec OpenAI et Anthropic

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

Related Stories

Jamie Oliver's Chewing Gum Ban Comment Resonates with Singaporeans

Jetstar Introduces New Size-Based Carry-On Baggage Rules

Fed Official Highlights Need to Monitor AI Buildout Finances

Gilead Projects Strong 2026 Sales and Increased HIV Growth Outlook

PERSPECTA