We’re entering a new age of AI moderation, but it may be too late to rein in the chatbot beast
Remember when AI was cool and everything was awesome? Of course not. AI is still cool and kind of awesome, but also occasionally problematic, and now we’re quickly moving into the amelioration and moderation phase. This is the moment where, after delivering untold AI companions and friendly models that could chat you up like a best friend, companies like Meta and OpenAI are backpedaling a bit.
The signs are everywhere, from a GPT model that was so staid and careful that it bored people to tears to a raft of Meta social media controls for teens aimed, at least in part, at curbing the AI companion access the company once touted.
In 2023, Meta rolled out AI chatbots for celebrities and cultural influencers like Snoop Dogg, Tom Brady, Kendall Jenner, and Naomi Osaka. Those were the marquee ones, but in reality, Instagram is now filled with an almost unlimited number of AI companions who will do everything from tell your fortune to spit out on-demand AI imagery.
It’s not that people need an AI face to make the connection. Earlier this summer, a Common Sense Media study found that 70% of teens in their survey reported using AI companions. And there are numerous reports of people falling in love with their AI chatbots.
Soon after the Common Sense Media study arrived, came the tragic news that a teen committed suicide after apparent “encouragement from ChatGPT.”
OpenAI’s far more cautious and less sycophantic GPT-5 arrived that same month and signaled a turn. It featured fewer hallucinations and more thoughtfulness in responses. In fact, it was slower to better consider its response, perhaps to avoid telling you or anyone else what you wanted to hear.
Instagram’s plan
More recently, Meta has pumped the brakes on teen access to AI content, at least not without parental oversight. Teens, by the way, are those between 13 and 18 years old. No one younger than 13 is supposed to be on these platforms in the first place. (Please hold for laughter.)
Instagram’s been pushing out parental controls for over a year, and these most recent ones that won’t arrive until 2026, offer perhaps the strongest set of AI controls yet, including the ability to totally block AI chatbot access entirely or just access to certain characters.
It feels like a start, but there are also signs in the AI industry that some want to operate on two, potentially conflicting planes at once.
In OpenAI CEO Sam Altman’s recent X post regarding upcoming changes to GPT-5, he claimed they had the company had made ChatGPT extra restrictive to protect those with mental health issues. “Now that we have been able to mitigate the serious mental health issues and have new tools, we are going to be able to safely relax the restrictions in most cases.”
We made ChatGPT pretty restrictive to make sure we were being careful with mental health issues. We realize this made it less useful/enjoyable to many users who had no mental health problems, but given the seriousness of the issue we wanted to get this right.Now that we have…October 14, 2025
He followed by saying that they would open ChatGPT and “treat adults like adults” and even offer the ability to generate “erotica for verified adults.”
So, yes, let’s protect vulnerable people (which I assume includes teens) while also opening up the platform to more adult content, just the kind of thing that might draw in…wait for it…more teens.
The claim that OpenAI has solved anything is pretty rich. It’s been clear from the start that virtually no one who is in the generative model training and building space actually knows before they launch these chatbots how real people will interact with them.
Altman, in particular, is the king of Launch Fast and Clean Up Later.
Meta, ostensibly, has with its vast stores of telemetry from billions of social media users, more knowledge than most about how people might respond to new tools. It took a long time, though, for it to see some of the harms social media was having on teens, or at least talk publicly about it.
By comparison, it’s taken far less time for Meta to introduce teen controls ranging from the promise of PG-13-grade AI content to completely blocking teens from some AI access.
Too little, too late?
I worry, though, that these tools come too late. A generation of teens is now as surrounded by AI tools and content as they were a decade ago by social media.
OpenAI talks about verified adults, and Meta says it will detect teen behavior in users and move them into the controlled space, but from what I know of teens (I had a couple myself), that all may do little to stop teens from accessing AI chatbots and trying to make them their digital friends.
There are no easy answers here, but if there’s one thing that comforts me, it’s the knowledge that, because we now live on AI time, the solutions will come faster than before. AI will become inherently safer, and maybe the age verification systems will, thanks to AI, become unbeatable.
Or not. Either way, this is one Pandora’s Box that we’re way past closing.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.
You might also like
Remember when AI was cool and everything was awesome? Of course not. AI is still cool and kind of awesome, but also occasionally problematic, and now we’re quickly moving into the amelioration and moderation phase. This is the moment where, after delivering untold AI companions and friendly models that could…
