OpenAI’s Child Exploitation Reports Increased Sharply This Year

OpenAI despatched 80 instances as many baby exploitation incident studies to the Nationwide Heart for Lacking & Exploited Youngsters through the first half of 2025 because it did throughout an analogous time interval in 2024, in keeping with a recent update from the corporate. The NCMEC’s CyberTipline is a Congressionally approved clearinghouse for reporting baby sexual abuse materials (CSAM) and different types of baby exploitation.

Firms are required by regulation to report apparent child exploitation to the CyberTipline. When an organization sends a report, NCMEC evaluations it after which forwards it to the suitable regulation enforcement company for investigation.

Statistics associated to NCMEC studies could be nuanced. Elevated studies can typically point out modifications in a platform’s automated moderation, or the standards it makes use of to determine whether or not a report is critical, moderately than essentially indicating a rise in nefarious exercise.

Moreover, the identical piece of content material could be the topic of a number of studies, and a single report could be about a number of items of content material. Some platforms, together with OpenAI, disclose the variety of each the studies and the overall items of content material they had been about for a extra full image.

OpenAI spokesperson Gaby Raila stated in a press release that the corporate made investments towards the tip of 2024 “to extend [its] capability to evaluation and motion studies with a purpose to hold tempo with present and future person progress.” Raila additionally stated that the time-frame corresponds to “the introduction of extra product surfaces that allowed picture uploads and the rising recognition of our merchandise, which contributed to the rise in studies.” In August, Nick Turley, vp and head of ChatGPT, announced that the app had 4 instances the quantity of weekly energetic customers than it did the yr earlier than.

In the course of the first half of 2025, the variety of CyberTipline studies OpenAI despatched was roughly the identical as the quantity of content material OpenAI despatched the studies about—75,027 in comparison with 74,559. Within the first half of 2024, it despatched 947 CyberTipline studies about 3,252 items of content material. Each the variety of studies and items of content material the studies noticed a marked enhance between the 2 time durations.

Content material, on this context, might imply a number of issues. OpenAI has stated that it reports all situations of CSAM, together with uploads and requests, to NCMEC. Moreover its ChatGPT app, which permits customers to add recordsdata—together with photos—and might generate textual content and pictures in response, OpenAI additionally affords entry to its fashions through API entry. The latest NCMEC rely wouldn’t embrace any studies associated to video-generation app Sora, as its September launch was after the time-frame coated by the replace.

The spike in studies follows an analogous sample to what NCMEC has noticed on the CyberTipline extra broadly with the rise of generative AI. The middle’s analysis of all CyberTipline information discovered that studies involving generative AI noticed a 1,325 % enhance between 2023 and 2024. NCMEC has not but launched 2025 information, and whereas different massive AI labs like Google publish statistics in regards to the NCMEC studies they’ve made, they don’t specify what proportion of these studies are AI-related.