Skip to content

Topic

AI Companions and Chatbot Relationships

AI chatbots designed to simulate friendship, romance, or emotional support, and the social, psychological, and safety effects of using them.

Current stories

leadershipConfirmed25 publishers

Disney sets a $20 million opening bid for Character.AI's IP in Chapter 11

Disney will open bidding at $20 million for a license to Character.AI's IP after the chatbot startup filed for Chapter 11 in Delaware. Character.AI says it will wind down the service if the sale produces no bid sufficient to keep the platform running.

Perspective Coverage

28 publishers
Builder
Builder 18%
Operator
Operator 34%
Investor
Investor 48%

Reality

Evidence78
Adoption
Not scored
Hype gap+15
Incentives60
Confidence80
productOne report1 publisher

Hot Girl Hotline charges $9.99 a month for AI-driven relationship advice, not companionship

Hot Girl Hotline is charging young women $9.99 a month for AI dating advice from a bot built to stop the conversation once it has helped. The startup is betting that people who already ask ChatGPT about relationships will pay for a bot that pushes them to act offline.

Reality

Evidence28
Adoption10
Hype gap+20
Incentives60
Confidence35
scienceOne report1 publisher

A Nature Human Behaviour framework models AI companion harm as four safety layers failing at once

Researchers at Singapore Management University and Duke-NUS argue that AI companions supplement people who already have social support and substitute for people who do not. They call governance the weakest of the four protective layers they name.

Publishers:phys.org

Reality

Evidence40
Adoption
Not scored
Hype gap+30
Incentives60
Confidence45
securityOne report1 publisher

AFT deal puts Microsoft's classroom AI under third-party privacy audits from November 1

The American Federation of Teachers got Microsoft to bar training on student data, ban companion features built for emotional attachment and accept outside audits. District buyers now have a written term sheet to put to every other vendor.

Publishers:securityweek.com

Reality

Evidence58
Adoption38
Hype gap+30
Incentives65
Confidence55
scienceOne report1 publisher

Negative posts nearly tripled on Replika's subreddit after an unannounced update

Harvard and Alberta researchers coded 54,861 subreddit posts around two real AI updates and surveyed 1,452 people. The reaction to Replika's silent change was larger, and it lasted longer than the reaction to GPT-5.

Publishers:phys.org

Reality

Evidence58
Adoption50
Hype gap+12
Incentives60
Confidence55
scienceOne report1 publisher

MIT's Sherry Turkle argues chatbots retrain what people expect from people

Her forthcoming book Artificial Intimacy argues that children especially will lose their reference point for real warmth by practicing on systems that only simulate it. The case is ethnographic, built from four decades of interviews.

Reality

Evidence45
Adoption
Not scored
Hype gap+25
Incentives60
Confidence60
investOne report1 publisher

Pleasant, and lonelier: a 12,365-person trial cuts against the AI companion pitch

A CESifo working paper randomised 12,365 French adults into four weeks of chatbot conversation. The talks rated more pleasant, while loneliness, life satisfaction and depressive affect all moved the wrong way.

Publishers:fortune.com

Reality

Evidence58
Adoption24
Hype gap+42
Incentives55
Confidence52