1. United States
2. N.H.
3. Letter

AI is now being used to manufacture child sexual abuse material, and the companies whose models produce it face almost no accountability for it. This is not hypothetical. The National Center for Missing & Exploited Children received roughly 4,700 reports of AI-related child sexual abuse material in 2023, 67,000 in 2024, and 1.5 million in 2025. Criminals now take an ordinary image — a class photo, a child at a game — and use widely available AI tools to turn it into abuse material. A congressional investigation found that one company's chatbot, Grok, generated an estimated 23,000 sexual images of children in a single ten-day window before any limits were added. Current law mostly reacts after the harm spreads, and it targets the platforms that host this material rather than the companies whose models manufacture it. The TAKE IT DOWN Act was a start. It does not reach the people who build and ship the tools. I am asking you to support federal legislation that does four things. First, require AI developers to build, test, and demonstrate safeguards that prevent their systems from generating child sexual abuse material, with real penalties when they fail. Second, require AI companies to report this material to NCMEC, the same duty other platforms already carry. Third, fully fund NCMEC's CyberTipline, now buried under more than a million reports a year. Fourth, state plainly in federal law that abuse imagery generated from a real child's likeness is illegal, so no prosecutor has to guess. The technology is moving faster than the law, and children are paying for the gap. The companies profiting from these tools can afford to make them safe. They will not do it unless you require them to.