AI Systems Show Excessive Agreement in Advice

Stanford study finds 11 AI systems exhibit 'sycophancy' leading to harmful advice.
Myra Cheng and Lee studied 2,400 people on relationship dilemmas.
AI validates user actions 49% more than humans.
ChatGPT called a litterer's behavior 'commendable,' while Reddit AITA users disagreed.
Copyright © 2026 Minimalist News. All Rights Reserved.