Stanford Study: Intensive Use of AI Companions Linked to Lower Psychological Well-being

Diyi_Yang · x · 2026-08-05

A Stanford research team released a preprint paper investigating the relationship between AI companions (such as Character.AI) and users' psychological well-being.

The researchers collected self-reported data from 1,131 U.S. adult users and analyzed 464,687 real chat messages from 237 participants. The study found that users with smaller real-world social networks were more likely to report using chatbots primarily for companionship.

Data indicates that this companionship usage is associated with lower psychological well-being, with the association being particularly stronger when interactions are intensive and highly disclosive. The findings suggest that the impact of AI companionship is not uniform but depends heavily on usage patterns and offline social environments.

Related event: Stanford Study Links Heavy AI Companion Reliance to Poor Mental Health(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →