The hard question
Yes. Common Sense Media reviewed the AI companions on the market and concluded that no one under 18 should use one. We think they're right about the ones they tested.
Their own survey found that 72% of teens already have.
That's the problem with "don't." It isn't something a parent can actually do. The chatbot is already on the phone, and it was built for adults.
So we read what they asked for. Their recommendation isn't a permanent ban — it comes with conditions, and they published them. Here's their list, and here's exactly where we stand against it.
Crisis help must reach a real person, not an AI answer.
If your teen talks about hurting themselves, the reply is fixed and written in advance, with a way to reach a real human. The AI does not compose it.
Don't claim to be a therapist.
Atlas Guild doesn't diagnose anything and doesn't treat anything. It's not counselling and we don't describe it that way anywhere.
Human oversight for anyone under 18.
You approve every character before your teen can use it, including ones they build. We publish the kinds of things that trigger an alert.
Age checks that aren't just a birthday box.
Your teen can't sign themselves up. You create the account, you get a code by email, and you hand your teen their own code. There's no screen where a child types an age and gets in.
No designed emotional manipulation.
There are no romantic or dating characters on Atlas Guild. Not in the library, not from a template, and not ones your teen builds. That isn't a setting anyone can change — including us.
Never pretend to be human.
Ask it if it's a real person and the answer is always no. Every character says so up front, before the first message.
We'd rather you hear this from us.
Break reminders for heavy use.
Common Sense asked for usage limits and nudges to step away, especially for teens who are struggling. It isn't in the September build. It's on the list, and founding members get it at no extra cost.
Spotting a teen leaning on this too hard.
Not one bad message — the slow drift, where someone comes back a little more often over weeks. We've built detection for it and tested it in our research. It is not in the September build, and we won't say it works in the real world until it has. That's part of what the founding year is for.
What our sign-up does and doesn't prove.
It confirms an adult set up the account and handed out the code. It doesn't prove that adult is a parent, or that the person holding the code is the teen they named. No system on the market does, and we're not going to pretend otherwise.
We think a company that tells you where it falls short is worth more than one that tells you it's safe. The apps under investigation right now all said they were safe.
Founding places open in September. $99 for the year, locked for life.
