Federal Appeals Court Rules Private Possession of AI-Generated CSAM Protected by First Amendment

A US federal appeals court ruled on 25 August 2026 in United States v Steven Anderegg that the government could not criminalise private possession in the home of obscene AI-generated child sexual abuse material where no real child was depicted. Judge John Z Lee of the Seventh Circuit Court of Appeals held that Supreme Court precedent from 1969 and 2002 required the outcome, though he noted the court had concerns about the lines those decisions set given current technological advances. Anderegg had used Stable Diffusion to generate photorealistic sexual images depicting prepubescent children; the government accepted that the material did not depict any actual child, and charges of production and possession were dismissed on First Amendment grounds, while charges of distribution and transfer to a minor under 16 remained.

Platform-level deterrence operates differently. Detection modules tuned to grooming escalation, image solicitation and coercive pressure do not distinguish between authentic and synthetic imagery in determining whether a conversation represents a threat to a child; they flag the pattern of sustained targeting, requests for increasingly explicit content, and threat language that characterise both traditional CSAM offences and AI-enabled abuse. By surfacing a child in distress to a parent or safeguarding professional as the exchange develops, systems such as Guardii close the operational gap the Anderegg ruling exposes, intercepting harm before material is possessed, stored or becomes the subject of a constitutional challenge and enabling rapid escalation to the appropriate authority when distribution or contact with a minor is detected.