Research02 · Harm
When AI use starts to hurt
Questions we ask
What happens when people hand their judgment to a model?
We study what happens when people stop deciding for themselves, and how they can stay the authors of their own choices.
Related concerns
Everyday judgment leaving the person
63 / 75 · miss -12
We mean the quiet, everyday habit of asking a model what is true, what to buy, what to say to a partner, or what a symptom means.
OpenParasocial AI as primary confidant
88 / 100 · miss -12
Loneliness plus always-on chat creates a soft substitute for human care: affirmation-seeking sessions that stretch for hours.
Open
From the library
Continuous read
Sourced measurements
Published measurements tied to concerns in this research area.
Companionship · perception · tier 2 · 100 pts
US states with enacted laws regulating AI companion chatbots or minors' chatbot use
18
US states · published 2026-09-01
“Thirty seven states introduced 124 chatbot-related bills in 2026, with thirteen of those signed into law–bringing the total number of state chatbot laws enacted in 2025 and 2026 to eighteen”
Future of Privacy Forum — Regulating the Chatbot: U.S. Chatbot Regulations Report
Count of enacted state chatbot laws 2025–2026; includes companion and minor-safety chatbot statutes.
Judgment · load · tier 1 · 75 pts
Share of US adults who use AI chatbots for health information or advice at least monthly
20%
US adults · published 2026-06-11
“To get medical advice — 20%”
Pew reports ever-use for medical advice among U.S. adults; panel asks at-least-monthly — treat as upper-bound load until a monthly cut publishes.
Companionship · load · tier 1 · 50 pts
Share of US adults who have used AI for emotional support or companionship
10%
US adults · published 2026-06-11
“For emotional support or advice — 10%”
Emotional support/advice is the closest national adult companionship measure; companionship alone is 4% in the same chart.
Judgment · perception · tier 1 · 75 pts
Share of US adults who say AI will worsen people's ability to think
53%
US adults · published 2025-09-17
“53% say AI will worsen people’s ability to think creatively, compared with 16% who say it will improve this.”
Pew Research Center — How Americans View AI and Its Impact on People and Society
Closest national measure to ability-to-think concern; wording is think creatively.
Companionship · load · tier 2 · 100 pts
Share of US teens who have used AI companions
72%
US teens ages 13-17 · published 2025-07-16
“Seventy-two percent of teens have used AI companions at least once, and over half (52%) qualify as regular users who interact with these platforms at least a few times a month.”
Common Sense Media — Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions
Companionship · load · tier 2 · 100 pts
Share of US teens who use AI companions regularly
52%
US teens ages 13-17 · published 2025-07-16
“Seventy-two percent of teens have used AI companions at least once, and over half (52%) qualify as regular users who interact with these platforms at least a few times a month.”
Common Sense Media — Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions
Companionship · load · tier 2 · 100 pts
Share of teen companion users who find AI conversations as satisfying as, or more than, real friends
31%
US teens ages 13-17 · published 2025-07-16
“Thirty-one percent of teens find conversations with AI companions as satisfying or more satisfying than those with real-life friends (21% say about the same quality, and 10% say more satisfying).”
Common Sense Media — Talk, Trust and Trade-Offs: How and Why Teens Use AI Companions
Reported among teens surveyed; companion ever-use is 72%.
Judgment · load · tier 1 · 50 pts
Share of US adults who have used ChatGPT or another AI chatbot
34%
US adults · published 2025-06-25
“Today, 34% of U.S. adults say they have ever used ChatGPT, according to a Pew Research Center survey.”
Pew Research Center — 34% of U.S. adults have used ChatGPT, about double the share in 2023
Panel asks ChatGPT or another chatbot; Pew measures ChatGPT specifically — a floor on broader chatbot use.