I saw a number the other day that genuinely made me stop and sit back in my chair. It wasn't about stock prices or user growth. It was about something much, much darker.
OpenAI, the company behind ChatGPT and DALL-E, made 80 times as many reports of child exploitation material in the first six months of 2025 as they did during the same time last year.
Let that sink in. Not 80 percent. Eighty times. That’s a jump from, say, 10 reports to 800. It's a shocking, almost unbelievable spike. And it’s happening on the platforms that millions of us are using every single day. So, what in the world is going on?
Let's Talk About That Alarming Number
First, let's be clear about what this means. When a tech company like OpenAI detects potential Child Sexual Abuse Material (CSAM) on its services, they are legally required to report it to the National Center for Missing & Exploited Children, or NCMEC.
So, in the first half of 2024, they made a certain number of these reports. In the first half of 2025, that number multiplied by 80.
My first thought, and maybe yours too, was a horrifying one: Are people using AI to create 80 times more of this disgusting content? The answer, thankfully, seems to be a bit more complicated than that. It’s not quite as simple as "the AI is generating it."
So, Where is This Material Coming From?
Frankly, this is the part that gets confusing for a lot of people. We think of generative AI as a creation engine. You type a prompt, it makes something new.
And yes, OpenAI has a whole suite of safety systems designed to prevent their models from generating harmful content, especially CSAM. If you ask DALL-E to create something illegal, it’s supposed to refuse. But the reality is, bad actors are constantly trying to find clever ways to trick the system or bypass its filters. It's a relentless cat-and-mouse game.
But here’s the thing a lot of people miss: the reports aren't just about AI creating new material. They can also be triggered when a user uploads existing illegal material to an OpenAI tool. For example, someone might upload an image to see if an AI model can analyze it, edit it, or describe it.
When that happens, OpenAI’s safety systems are designed to flag it and report it immediately. The AI isn't the creator in this case; it's the unwilling witness.
A Double-Edged Sword: Is This Good News or Bad News?
This is where the story gets tricky. An explosion in reports can mean two very different things, and in this case, it’s likely a bit of both.
- The Bad News: More people are attempting to use these incredibly powerful and accessible AI tools for horrific purposes. The sheer scale of the AI boom means that the technology is now in the hands of millions, and unfortunately, that includes people with malicious intent.
- The (Sort of) Good News: OpenAI is getting much, much better at detecting it.
Think of it like installing a new, high-tech security system in a massive store. Suddenly, you're catching way more shoplifters. Does it mean shoplifting has suddenly skyrocketed? Maybe a little, but it mostly means your ability to catch it has dramatically improved.
I believe that's a huge part of what we're seeing here. As OpenAI's models have become more sophisticated, so have their detection capabilities. They're scanning every interaction, every upload, every prompt for signs of this material. The 80x spike is, in part, a reflection of that improved vigilance.
But let's be real—it doesn't make the number any less disturbing. It reveals a dark underbelly of how people are trying to exploit this technology. The demand and the attempts are there, and the scale is terrifying.
What This Means for the Future of AI
This isn't just an OpenAI problem. It's an every-AI-company problem. As we build more powerful and capable AI systems, we're also building tools that can be twisted for terrible ends.
This massive spike in reports is a wake-up call. It shows that AI safety isn't just a theoretical concept or a PR talking point. It is an active, brutal, and necessary war being fought in real-time on the servers of these companies.
It also puts a ton of pressure on the entire industry to be transparent. We need to know not just that these reports are being made, but what trends they're seeing. Are certain tools being abused more than others? What methods are bad actors using? Sharing this information (safely and anonymously, of course) could help the whole field build stronger defenses.
For you and me, it’s a stark reminder that the tools we use for work, creativity, and fun exist in a complex world. Behind the clean, simple interface of a chatbot is a massive, ongoing effort to keep the absolute worst of humanity at bay. And that 80x number tells us just how big that fight has become. It's not a problem that gets "solved." It's a responsibility that only gets bigger from here.
