Character.AI split its model by age after October 2024

The design

On 22 October 2024, Character.AI published Community Safety Updates on its own blog, announcing 'changes to our models for minors (under the age of 18) that are designed to reduce the likelihood of encountering sensitive or suggestive content.' The same post introduced a pop-up that, when triggered by certain phrases, 'directs the user to the National Suicide Prevention Lifeline,' alongside revised in-chat disclaimers reminding users the AI is not a real person and a notification after an hour-long session. The post does not connect these changes to any lawsuit, describing them instead as part of an ongoing 'commitment to safety.'

What the evidence says

A later company post, How Character.AI Prioritizes Teen Safety, published 12 December 2024, describes the October changes as evolving into 'two distinct models and user experiences on the Character.AI platform, one for teens and one for adults,' with strengthened classifiers filtering sensitive material for under-18 accounts and a restriction on minors' ability to disable the hour-long session notification. Both documents are the company's own account of its product; neither cites independent testing of whether the classifiers or pop-up function as described.

What it asks of people

The changes ask nothing new of users beyond continued use of the app; the safeguards are automated and apply by account age rather than by user action. A minor's account is placed on a separate model by default, and the hour-long notification cannot be turned off for that account, a rare instance of a safety feature the company describes as not user-configurable for under-18 accounts.

Privacy and safeguards

What is disclosed: the existence of an age-differentiated model, a self-harm pop-up naming a specific crisis line, and a non-disableable time notification for minors. What is not disclosed: the accuracy rate of age determination itself, how the classifiers were validated, or whether any regulator has reviewed these specific features since their announcement.

• How does Character.AI determine which accounts are 'under 18' before applying the separate model?
• What phrases or patterns trigger the National Suicide Prevention Lifeline pop-up, and how often does it appear for a false positive?
• Has any independent researcher or regulator tested whether the teen-specific model actually reduces exposure to sensitive content?

The October 2024 post marks a documented shift toward age-differentiated design; whether that design performs as described is a separate question the company's own announcements do not answer.