Ed Ludlow

In a statement, an Amazon spokesperson said the training data was from external sources and the company doesn't have the details about its origin that could aid investigators.

Bloomberg's Riley Griffin reported on and broke this story.

I think let's start with the very basics of what this entity that we reported with saw in the data when it came to specifically AI-related reports of child sexual abuse material.

Just start with that, please.

Yeah, so there's an organization called the National Center for Missing and Exploited Children, and it effectively serves as a clearinghouse, fielding tips from industry of child sexual abuse material to law enforcement.

This is a really important part of the process because this is the connectivity that allows local law enforcement agents to actually track down children who might be in active danger and find the perpetrators of abuse.

They field these reports from companies like Amazon.

And when hundreds of thousands of reports came in, they typically expect details, location data, sources.

Where did you find this data?

And Amazon has not shared that detail, which has stunted further investigation.

Is that why Amazon's data is seen as some sort of outlier here, Riley?

Yeah, so NCMEC, the clearinghouse, has been tracking this broader bucket of AI-related reports.

That can be AI-generated child sexual abuse material, exploitive conversation with chatbots.

It could also be known images of abuse that have existed on the Internet.

And ultimately, this bucket is rising.

This year, 2025, was the first in which they saw one million reports of AI-related child sexual abuse material.

And Amazon drove the vast majority, so hundreds of thousands of those cases.

That's what makes it an outlier here, in addition to the lack of information.

This is a story about how artificial intelligence is changing the child safety landscape.

And in the course of reporting, you spoke with a wide range of experts in this field.