AI Child Abuse Images Jump 20-Fold as Hacking Tools Sell for 10 Cents | #hacking | #cybersecurity | #infosec | #comptia | #pentest | #hacker

Open-source artificial intelligence models are being exploited for sex crimes. AI-generated child sexual abuse material, or CSAM, has increased more than 20-fold in a year, driven in part by prices so low that AI hacking tools change hands for 10 cents on the dark web. The tools are cheap and difficult to trace.

Reports of AI-linked CSAM received by the U.S. National Center for Missing & Exploited Children rose from 4,700 in 2023 and 67,000 in 2024 to more than 1.5 million last year, Bloomberg reported on the 7th. Britain’s Internet Watch Foundation said it identified 6,310 realistic AI-generated child sexual abuse images in the first half of this year.

Users who download free software onto a personal computer can train it further on existing child abuse images and generate unlimited numbers of similar pictures. Because the program runs offline, prompts are never sent to a commercial service, leaving no opening for the activity to be flagged as illegal and reported to police. Commercial services carry safeguards against child sexual exploitation material, but free AI programs downloaded and run on personal computers bypass them.

A European investigator who asked not to be named said the number of models and variants under surveillance had grown from fewer than 12 two years ago to more than 500 now. Models cited by child safety experts and investigators who handle sex crimes against minors include Stable Diffusion from Stability AI of Britain and the U.S., Flux from Germany’s Black Forest Labs, Qwen from China’s Alibaba and Hunyuan from Tencent. Danny van Altuis, who leads a team at Europol, said that with AI there is no limit to the imagination, and that the open-source models used in these crimes look as though they were built for crime.

AI models are frequently trained on actual abuse images, sometimes using pictures of victims taken years or decades earlier. As a result, even fully synthetic images continue to feature real victims or figures resembling them in new abuse material, perpetuating a cycle of exploitation.

AI Hacking Tools for 10 Cents on the Dark Web, 1,400 Black Market Ads a Month

The same structure is at work in the cybercrime market. Axios, citing a report by the Halcyon Ransomware Research Center, reported that advertisements for AI tools posted on black market platforms surged from fewer than 50 a month late last year to more than 1,400 in February. Researchers analyzed about 4,000 posts across 77 Telegram channels, 20 dark web forums and five specialized black markets from October last year through May this year.

Sellers offered access to ChatGPT Plus accounts for 10 cents and jailbreak prompts that circumvent model safeguards for 32 cents. Prompts that strip the safeguards from major models traded from $10. Cynthia Kaiser, head of the Halcyon Ransomware Research Center, said the findings show how little it costs to carry out such activity and that this should frighten companies.

The AI tools being traded include a system capable of placing calls to as many as 120 people simultaneously. The report noted that criminals have fully commercialized both sales and production, often replicating software-as-a-service subscription models and online storefronts. Some sellers were found to move tools vetted on dark web forums to Telegram shops, which are easier to manage.