Meta AI now gives you insight into your child’s conversations with its AI chatbot. But this feature only shows broad strokes, and that’s not enough to meaningfully monitor and protect your child.
Meta AI has been ringing alarm bells since its inception. It earned an “unacceptable” risk assessment by Common Sense Media, faced an FTC investigation, and even failed Meta’s own internal testing.
Here are some of the issues with it:
Recently, Meta blocked underage users from interacting with AI characters and announced that teens inquiring about suicide will trigger a parental alert. These sound promising, but I’m still skeptical.
Blocking interaction with AI characters is good, but this doesn’t address the entirety of the problem. Teens can still use Meta’s AI assistant, where many safety concerns remain.
Additionally, no independent tests of the new notification system have been published, and Meta hasn’t released internal testing data. Given that third-party stress tests of Meta AI’s prior safety notification system declared it “broken” and notification systems from competitors like ChatGPT are also resounding failures, I’m not encouraged that this round is anything more than empty promises.
If you supervise your child on a Teen Account on Facebook, Messenger, or Instagram, you can now see what they’ve asked Meta AI.
Via the Insights tab, you’ll see a list of the topics your teen has asked about in the past seven days. (Your teen can also see the same list.)
Topics are broken down into 16 categories:
If you click on a category, you get associated sub-topics, including at least one your child has asked about. However, you aren’t able to see the exact prompt your child entered, nor the chatbot’s response.
Meta does provide 11 conversation-starters developed in collaboration with the Cyberbullying Research Center to help you follow up with your teen. (It’s a great resource.)
Meta’s new Insights feature strikes me as more safety theater from a company that consistently prioritizes profits over child safety. Here’s why Meta’s “topic categories” don’t qualify as true supervision and aren’t enough to keep your child safe:
Expert warnings about AI chatbots range from potentially harmful to downright unsafe and unacceptable for users under 18. Instead of limiting use to adults or building a genuinely safe product, Meta made a subpar feature that implicitly signals to parents they should let their kids use AI chatbots because they’re now “safe.”
All you’re told is that something your child typed fits into one of 16 broad categories. Clicking for details only reveals a handful of sub-topics, some of which may not be related to what they asked. That’s not meaningful insight. For example, if Health and Wellbeing appears, your child might have asked about acne or about serious issues like mental health or substance abuse.
Although Meta’s list includes 16 topics, it inexplicably limits you to 10 items per seven-day period. This arbitrary cap means if your child’s 11th topic is something serious, you’re out of luck.
It’s tempting to say Meta’s Insights are better than nothing, but I’m not convinced. It may give a false sense of security and cause parents to forgo additional measures. I don’t know that I’d go so far as to say it would be better if they instead did nothing, but at least without this feature, it would be clearer that the platform doesn’t truly monitor kids, and parents could make safety decisions accordingly.
Users can’t fully opt out of Meta AI, but they can mute the chatbot. Given teens’ tendency to do the opposite of what parents ask, rather than telling them to mute it, provide them information on why it’s bad for them.
It may not be a big leap; according to my 14-year-old niece, “teens hate AI,” and she’s not alone in her sentiment. This stance makes sense for a digitally savvy generation: the more educated people are about how AI works, the less likely they are to use it.
Demonstrate critical thinking when you use AI and make it known when you’re skeptical of its output.
BrightCanary scans everything your child types, including on all AI platforms, and gives you meaningful insight to help you address the situation:
AI chatbots are incredibly risky for kids. Meta AI’s Insights feature uses broad categories that aren’t a substitute for true parental monitoring and risk giving a sense of false security. BrightCanary gives you the visibility and details you need without leaving your child’s safety up to guesswork. Download today to get started for free.
