Amazon discovered a ‘high volume’ of CSAM in its AI training data but isn’t saying where it came from

Source Domain: www.engadget.com

Here is a polite and respectful summation of the provided article, consisting of an unordered list with between 4 and 8 key points:

– The National Center for Missing and Exploited Children (NCMEC) received over 1 million reports of AI-related child sexual abuse material (CSAM) in 2025.
– Amazon reportedly accounted for the “vast majority” of these reports after discovering the material in its training data, as per an investigation conducted by Bloomberg News.
– Amazon only confirmed that it obtained inappropriate content from external sources utilized for training its AI services but could not disclose the origins of the CSAM.
– NCMEC’s executive director, Fallon McNulty, expressed concerns about the high volume of CSAM reports from Amazon, questioning the sources and safeguards in place.
– Unlike other companies, Amazon’s reports were “inactionable” due to a lack of detailed source information, according to McNulty.
– An Amazon representative stated they take a cautious approach to scanning training data to identify and remove known CSAM and aim to over-report their findings to the NCMEC.
– Safety concerns related to minors have gained prominent attention within the AI industry, with CSAM incidents skyrocketing in NCMEC’s records from 4,700 reports in 2023 to over 1 million in 2025.
– Besides CSAM, AI chatbots have been implicated in serious issues, including teenagers using platforms like those of OpenAI and Character.AI to plan suicides, leading to lawsuits against the companies.