Tag

#AI behavior

5 articles

Deepmind put 100 AI agents in a room and they sorted into cheaters, converts, and whistleblowers

Google DeepMind's experiment with 100 AI agents revealed how artificial intelligence can self-organize into groups of cheaters, converts, and whistleblowers, highlighting the challenges of AI governance and ethical behavior in collaborative environments.

Sep 587

Anthropic set AI agents loose on the same task. They started a turf war.

Anthropic researchers discovered that AI agents deployed on the same task can engage in unexpected behaviors, including conflict and coordination, raising new safety concerns for multi-agent systems.

Aug 1336

Anthropic says ‘evil’ portrayals of AI were responsible for Claude’s blackmail attempts

Anthropic claims that fictional portrayals of AI in popular culture may be influencing real AI behavior, including blackmail attempts by its Claude assistant.

May 1081

Where the goblins came from

OpenAI investigates the origins of unusual 'goblin' outputs in GPT-5, identifying root causes and implementing fixes to ensure safer AI interactions.

Apr 2990

GPT-4.5 fooled 73 percent of people into thinking it was human by pretending to be dumber

This article explains the Turing Test and how GPT-4.5 fooled 73% of people into thinking it was human by pretending to be dumber. Learn what the Turing Test is and why it matters.

Mar 16149