“Six people were sentenced for creating pornographic materials using AI “escape” methods; the judge reminded that “technical neutrality ≠ technical innocence”

On September 8, 2026, the People’s Court of Yuyao City, Zhejiang Province, concluded a case involving the use of generative AI to create pornographic content for profit. The defendant Zheng and five others formed a team and broke through the security restrictions of large models by rewriting system prompts and setting up character cards, turning general AI tools into criminal instruments for generating illegal content. This illegally operated AI model was launched overseas in January 2025, generating over 370,000 static images and more than 6,000 dynamic images during that period. Approximately 9,000 static images and 5,000 dynamic images were identified as pornographic content, with user payments reaching $24,000. The court determined that the six individuals committed the crime of creating pornographic content for profit, stating that “technical neutrality” is not a shield against criminal liability. Developers who actively tamper with model security mechanisms and have substantial control over the generation of illegal content should bear criminal responsibility.