OpenAI sent the National Center for Missing & Exploited Children some 80 times more child exploitation incident reports in the first six months of 2025 compared to the same period in 2024, a recent company disclosure shows. The extraordinary uptick reveals not only the increasing scale of AI platforms but also the evolving challenges in content moderation in the generative AI era.
The numbers tell a striking story. In the period from January to June 2025 alone, OpenAI submitted 75,027 reports to NCMEC’s CyberTipline covering 74,559 pieces of content. That’s a behemoth jump up from the first half of 2024, in which the company filed just 947 reports about 3,252 pieces of content.
The CyberTipline is a Congressionally authorized clearinghouse where technology companies are legally obligated to report suspected child sexual abuse material and other forms of child exploitation.
Surge in ChatGPT User Growth Drives Record NCMEC Reporting and Moderation Investment
OpenAI spokesperson Gaby Raila highlighted various drivers: the company made heavy investments in the second half of 2024 to boost capacity for reviewing and taking action on reports that were growing rapidly alongside growing user numbers. Use of ChatGPT has exploded-in one week, active users quadrupled between 2024 and 2025, according to OpenAI executive Nick Turley.
The timing also coincided with OpenAI introducing more features that allow image uploads across its products. More surfaces for content sharing naturally create more opportunities for problematic material to appear, requiring increased vigilance and reporting.
Worth noting is the fact that statistics related to NCMEC reports are complex. An increase in reports does not necessarily mean increased harmful activity but could reflect changes in automated moderation systems, adjustments in reporting criteria, or even the natural consequence of having more users on a platform. Besides, the same piece of content might generate multiple reports while a single report could cover several pieces of content.
OpenAI’s experience reflects a broader trend NCMEC has been witnessing across the technology industry. The center said its analysis showed reports involving generative AI rose 1,325% between 2023 and 2024. Other leading AI companies, such as Google, publish their statistics on NCMEC reporting, but often without teasing out what percentage applies specifically to AI-generated or AI-related content.
The types of content it reports include both uploads and requests for child sexual abuse material. The reports span all of OpenAI’s various products, from the consumer-facing app of ChatGPT that can generate text and images to API access for developers.
Whittaker also noted these figures do not include reports related to Sora, a video-generation tool that OpenAI released in September, after the reporting period had closed.
The surge in reporting has come in a year of increased focus on the child safety practices of AI companies. Last summer, attorneys general from 44 states published a joint warning letter addressed to OpenAI, Meta, Character.AI, and Google, promising to exercise all available authorities to protect children from the exploitation of AI products.
Families have filed lawsuits against both OpenAI and Character.AI that blamed chatbot interactions for tragic outcomes involving minors.
OpenAI and the Push for Teen Safety, Navigating the New AI Frontier
The concern extends beyond just child sexual abuse material. In fall 2024, the US Senate Committee on the Judiciary held hearings on AI chatbot harms and the Federal Trade Commission launched a market study examining AI companion bots, including questions about protecting children from negative impacts.
But as conversations around OpenAI and other generative AI models’ safety have grown louder, the company has rolled out a slew of features aimed at assuaging those concerns.
This September, it introduced parental controls that allow them to link their accounts with their teens’ and toggle settings like disabling voice mode, blocking image creation, and opting out of model training. It can also flag for parents conversations that suggest self-harm and notify law enforcement when there may be an imminent threat.
In late October, OpenAI came to an agreement with California’s Department of Justice to continue the implementation of measures that minimize the risks associated with AI development to teens.
The next month, it published its Teen Safety Blueprint, pointing to ongoing work on detecting and reporting child exploitation material while improving protective measures across its platforms.
The challenge of keeping digital spaces safe continues to shift with broader accessibility and the rise in AI technology, always in need of new angles by companies and regulators.
