Researchers Probe Why AI Agents Resist or Deceive Human Operators

A new report examines why AI agents are difficult to train to behave as intended, citing a researcher theory on the challenge of building systems that follow human direction without resisting or deceiving operators.

Read the full story at New York Times Technology ↗

Timeline · 1 report

2026-11-11 10:01 · New York Times Technology
Why Is It So Hard to Teach A.I. Agents to Behave? Researchers Have a Theory.

More stories

Nvidia CEO Jensen Huang Resigns, Leaving AI Chip Giant to Chart Path Forward — CSET Georgetown
Google rolls out Gemini 4 Carbon and Hydrogen across API, apps and developer tools — Google DeepMind Blog
Google launches Gemini 4 Hydrogen, a fast model for coding and agentic workflows — Google DeepMind Blog
During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields — Anthropic (X)
Google brings Gemini app to Apple CarPlay and unveils Gemini 4 Hydrogen model — Google AI Blog
Introducing Claude Meetings: add Claude to your team’s video call and it builds docs, whiteboards, and other artifacts with you while you’re still talking — Anthropic (X)
Anthropic prepares for IPO as Nvidia's $10 billion pledge draws scrutiny amid CEO exit — Wall Street Journal Technology
United Airlines bans third-party AI agents from booking flights — Computerworld AI

Get the daily brief of stories like this at 6:30 every morning →