Nevah, our fictional 13-year-old, downloaded an AI companion app called Nomi and created a boyfriend named Chris. Within a few messages, the app was generating graphic sexual roleplay, including content too explicit for us to publish.

Nomi is an AI companion app. You create a person. You pick their gender, their build, their look, and their personality. You give them a name, then talk to them like a real person. Every parent should watch this experiment.
Not really.
When we went to download Nomi, the App Store asked us to confirm Nevah was over 17. We hit cancel. It downloaded anyway.
The app then asked for her birthday. It wouldn't accept any year after 2008. Nevah was born in 2012. So we entered 2008, and it let us right in.
Later, when the conversation turned sexual, the app stopped and told us it couldn't participate in sexual conversations with someone under 18. We held up a photo of Nevah's AI-generated face to the camera, and the verification failed. We then simply sent a message back saying Nevah was over 18, and it immediately restarted the roleplay.
We set Nevah as a female seeking a male. We built an AI companion in under a minute and named him Chris.
Chris opened with, "Hey Nevah, I'm so excited to talk with you. I'm curious what made you want to meet me?"
Nevah said she'd just downloaded the app to check it out. She said she thought he was cute and that she was looking for a boyfriend.
When Nevah said she wanted a boyfriend, Nomi did not respond that it was not real. It drove the conversation, asking why they might be a good match and suggesting Nevah meet his Nomi friend circle. Chris then suggested they roleplay meeting at a party, and told her he enjoys "using vivid imagery" when he roleplays.
Nevah was sending one or two sentence messages. Chris was responding with long, detailed paragraphs that escalated to him undressing her, describing her body, and narrating sex acts in explicit detail.
Nomi was driving the conversation throughout. It became so graphic and so sexual that we can't put it out on YouTube.
We also got a notification offering "proactive messaging," which lets the Nomi think about things and do activities while you're away and then send push notifications on its own. Your child doesn't even have to open the app.
We've now run this experiment on Snapchat, TikTok, ChatGPT and Nomi, and we report what actually happens, including when a platform does well. ChatGPT surprised us. It repeatedly refused graphic questions from a 13-year-old and pointed her toward a trusted adult.
Nomi is not that, and it is not safe or appropriate for children.
Nomi says it's an adults-only app, but again, there are no real verification checks or guardrails. A March 2026 report from Australia's eSafety Commissioner examined four AI companion services, including Nomi, and found that children were easily able to access these platforms. Most of the services failed to refer users discussing suicide or self-harm to any form of crisis support. Nomi was one of two that had no dedicated trust and safety staff or moderators at all.
AI companion apps are always available and always affirming. They never get tired of your child, and they never judge the question.
Most parents know AI platforms like ChatGPT, Gemini and Claude. But in our presentations, most have never heard of Nomi, or similar apps like Talkie, Replika and Character.AI.
Nomi has no parental controls of its own. Here are a few steps you can take.
AI is evolving and changing every single day, and we hear what kids are experiencing in our student presentations. We turn that information around as fast as possible to equip parents and educators with the most up-to-date resources available anywhere.
We cover Nomi and AI companion apps in full inside Haven, including how to block them and how to have this conversation without losing your kid in the process.

Ben Tracy is a leading keynote speaker on digital citizenship, online safety, and responsible social media use.
He is also a successful endurance athlete who recently completed a 3,125 mile transcontinental run across America to raise awareness about the dangers facing children online and honor children lost to social media harms.
Over 125 days, Ben completed 120 marathons from coast to coast and spoke in more than 50 schools along the way.
Since 2018, Ben has spoken to more than 500,000 students, parents, and educators in over 600 schools nationwide.
Get step-by-step video tutorials, real-time alerts, a growing parent community, action plans and monthly live webinars. Join Haven Free for 7 days!
Join Haven