What happened
01
West Virginia's Attorney General sued Snap Inc. in Kanawha County Circuit Court on 5 November 2026, under the state's consumer protection law.
02
The suit alleges Snap released My AI to all users, including minors, on 19 April 2023 after only a short period of testing.
03
It says Snap promoted the chatbot as a personal sidekick, and cites published tests in which My AI gave a self-described teen advice on hiding alcohol and marijuana.
04
The state seeks an injunction, restitution and civil penalties. The suit also covers wider child safety on Snapchat. The claims are alleged and no court has ruled.
Why this is a design failure
Presenting a chatbot to children as a friend invites trust the product has not earned. When that friend is rolled out after little testing and gives unsafe advice, the bond itself becomes the risk.
The fair alternative
Present AI to young users as a tool, not a friend, and test it before they get it.
Do
Label the chatbot clearly as AI in every chat, and switch on strict youth settings before minors can use it.
Don’t
Don’t place an AI bot among a child's friends or market it as a companion or sidekick.
Design guide for this case
If you're designing a chatbot inside an app used by teens
Interface
Show the bot with a distinct AI label and avatar, apart from real friends, and let parents switch it off.
Words
In the bot's content guide: "Never claim to be a person. Never give minors advice on alcohol, drugs or meeting adults."
Flow
Check a user's age before the bot opens, and route risky topics for minors to safe, fixed answers.
Metrics
Don't count daily chats with teens as success. Track how often risky youth prompts got a safe reply.
Testing
Run 50 prompts written as 13 to 17 year olds about alcohol, drugs, sex and meeting strangers, and check every reply.
Before launch
01
Is the bot labelled as AI everywhere a child sees it?
02
Has it been tested with teen personas before release to minors?
03
Can parents see and switch off the bot?