S01¶

WebSearch — Stanford sycophancy study confirming 49% AI affirmation rate

Summary¶

Field	Value
Source/Database	WebSearch
Query terms	AI models affirm users views 49% more often than humans sycophancy study 2024 2025
Filters	None
Results returned	10
Results selected	3
Results rejected	7

Result	Title	URL	Rationale
S01-R01	Stanford Report: AI overly affirms users	https://news.stanford.edu/stories/2026/03/ai-advice-sycophantic-models-research	Primary university press release
S01-R02	Science journal publication	https://www.science.org/doi/10.1126/science.aec8352	Peer-reviewed primary source
S01-R03	Fortune coverage	https://fortune.com/2026/03/31/ai-tech-sycophantic-regulations-openai-chatgpt-gemini-claude-anthropic-american-politics/	Major outlet confirming figures

Result	Title	URL	Rationale
S01-R04	Rappler AI sycophancy coverage	https://www.rappler.com/technology/features/ai-sycophancy-study-science-journal-march-2026/	Derivative coverage, no new information
S01-R05	Neuroscience News coverage	https://neurosciencenews.com/ai-sycophancy-moral-judgment-30397/	Derivative coverage
S01-R06	CapitalAI Daily coverage	https://www.capitalaidaily.com/stanford-study-finds-ai-models-agree-with-users-49-more-than-humans-with-harmful-effects/	Derivative coverage
S01-R07	Dagens coverage	https://www.dagens.com/news/ai-chatbots-may-be-making-users-worse-people-study-finds	Derivative coverage
S01-R08	H2S Media coverage	https://www.how2shout.com/news/stanford-study-ai-chatbots-sycophancy-agree-users-science-journal.html	Derivative coverage
S01-R09	MIT Sloan ME coverage	https://www.mitsloanme.com/article/when-ai-agrees-too-easily-it-may-be-changing-how-we-think/	Derivative coverage
S01-R10	KXAN coverage	https://www.kxan.com/news/national-news/self-affirmations-from-ai-chatbots-harm-human-relationships-study/	Local news derivative

All 10 results referenced the same Stanford study published in Science. The 49% figure was consistently reported across all sources.