Over the last few years, it has become accepted knowledge that the internet — mainly social media and, lately, AI tools — was damaging the mental health of kids and increasing the suicide rate among that demographic. To hear the media and politicians talk about it, this was the biggest issue on the planet and we had to do basically everything possible to block social media (and AI) from kids.
Of course, the narrative was always bullshit. While there had been an uptick in suicide rates among teens between 2010 and 2020, they were way below the rates in the 80s and 90s. And by the time Jonathan Haidt’s “moral panic in a book” came out blaming social media for teen mental health problems, the data already didn’t support his claims, as I pointed out in my review of the book right after it came out.
In that review of the book, I highlighted that while suicide rates among teens had ticked up in the US, they had actually declined in many other countries — including France, Spain, New Zealand, and elsewhere — despite plenty of social media usage in those countries. Similarly, the data on mental health issues was clouded by (1) much more openness to talking about mental health, and (2) importantly, new guidance under the Affordable Care Act that mandated increased screening for depression in adolescent girls, while simultaneously instructing clinicians to record suicidal ideation differently than in the past, which resulted in a massive uptick in such numbers.
And now the narrative violation continues, as the latest CDC survey data shows that mental health concerns among teens have been declining since their 2021 peak (i.e., as we were coming out of the worst of the COVID crisis).
Continuing a trend first observed in 2023, teenagers surveyed at high schools across the country in 2025 reported improvements in several key measures, including persistent feelings of sadness or hopelessness, serious thoughts of suicide and attempting suicide in the last year.
Some might argue that this is due to the work of Haidt and politicians banning social media for kids, but the data doesn’t seem to show that at all.
Thirty-six percent of teenagers said that they used social media every hour or more, a similar percentage to what was seen in 2023, when the data was first reported.
And while this survey doesn’t track AI usage, we know from other research that a large majority of teens now use AI chatbots, with roughly three in ten using them daily.
This does not mean, of course, that we can declare victory. There are still, clearly, plenty of teens dealing with mental health challenges.
But the data show that many teenagers, especially girls and teenagers who identify as lesbian, gay or bisexual, are still struggling. Thirty-three percent of high school students reported persistent feelings of sadness or hopelessness in the last 12 months, down from a peak of 42 percent in 2021. Fourteen percent of teenagers reported that they had seriously considered attempting suicide, down from 22 percent in 2021.
As actual experts have been pointing out for years, the reality around mental health and suicidal ideation is extremely complicated and not fully understood. But, over and over again it seems that the approach recommended by actual experts is not blanket bans on technology or simply “blaming” anything, but rather better efforts at identifying those going through mental health challenges while simultaneously making more mental health resources available to all who need them.
But, the narrative is the narrative. Hell, it was just weeks ago that a judge in New Mexico took it as irrefutably proven that social media was damaging to kids’ mental health. Does anyone get to go back and show him the latest data suggesting that maybe something else was driving this? Or do we just keep treating the narrative as fact?
This is what I’ve been saying for years: when we misdiagnose the problem and blame the wrong thing, the “solutions” we get are not solutions at all. Indeed, they could be making the problem worse by not actually putting in place the tools and resources to help those in need while, simultaneously, doing real damage to groups that rely on internet communities for their own mental health support.
I’d like to think that as more data like this comes out, further debunking Haidt’s narrative, we could get back to a serious discussion about how to actually help kids who are struggling. But I fear the easy comfort of a moral panic is too strong a pull for people to give it up that quickly.