A new peer-reviewed study suggests that artificial intelligence is reshaping teen sexual behavior in the US. The research, led by Chad Steel of George Mason University in PLOS One, surveyed 557 US teens aged 13–17. It found 55.3% had created at least one “nudified” image and 54.4% had received one. The authors describe this as a shift from traditional sexting to AI-enabled image fabrication.

The study characterizes today’s adolescents as “AI-natives,” not just digital natives. They are growing up with tools that can produce sexualized content from text prompts. These tools can “nudify” people without their consent, according to The Independent.

Substantial non-consensual activity

The findings point to substantial non-consensual activity. In the survey, 36.3% of teens reported that someone else had created a sexualized GenAI image of them without their consent. Another 33.2% said such an image of them had been distributed non-consensually. Researchers note that these practices can constitute sexual exploitation and often contravene US federal child pornography laws. Victims describe harms similar to those seen in other forms of child sexual exploitation material.

Rather than exchanging self-produced images in private, teens increasingly use accessible AI tools to create sexualized depictions of peers, acquaintances, or themselves. Many teens appear to view these practices as part of a new, technology-inflected social dynamic. Steel suggested that policymakers consider updating laws to distinguish consensual behavior from exploitative acts. He also notes that the non-consensual creation or sharing of sexualized images of minors remains illegal and harmful.

Offlimits, a Dutch foundation that tracks sexual boundary violations, has called for a ban on the “de-nuding” feature of Grok, an AI system. The group cites more than 250 reports in the past year of manipulated and AI-generated nude images, a nearly fivefold increase. OpenAI CEO Sam Altman is pushing to release an “adult version” of ChatGPT, and Elon Musk has advocated allowing Grok to generate R-rated content, according to Digital Trends.