How Maine is wading into the regulation of explicit AI-generated images of children

Child sexual abuse material is illegal under both state and federal law. It has been for decades. But the advancement of Artificial Intelligence created loopholes. 

That was shown over the summer when a man in Maine used AI to manipulate pictures he took of children playing in a soccer game into sexually explicit images. Police knew who the man was and what he was doing but couldn’t charge him with a crime under Maine law, as first reported by the Bangor Daily News. 

A proposal before the Maine Legislature seeks to amend the criminal code so that charges in such cases can be brought. 

The bill, LD 524 sponsored by Rep. Amy Kuhn (D-Falmouth), expands the crime of disseminating child sexual abuse material to apply to materials that have been created or modified through generative AI or machine learning. 

“It’s not a hypothetical anymore,” Kuhn said. “This technology has created new ways to victimize children, and our laws haven’t been updated to adequately protect against that.”

Last year, following concerns that it wasn’t constitutional, lawmakers passed a watered down version, LD 1944, that added the dissemination of “morphed images” as a form of harassment under the “revenge porn” law. But now a broader proposal to expand the definition of sexually explicit material is back, and lawmakers from both parties seem open to it.

During the first week of this year’s legislative session, the Judiciary Committee advanced that bill in a unanimous bipartisan vote. Tasked with grading the bill’s potential effects, on Wednesday the Criminal Justice and Public Safety Committee unanimously agreed the changes would result in a moderate impact on Maine’s criminal justice system. The chambers are expected to take votes in the coming weeks. 

So far, 28 states have banned the creation of AI-generated child sexual abuse material, according to MultiState, a government relations firm tracking state AI laws. That includes nearby New Hampshire, where on Jan. 1 the use of artificial intelligence to create “intimate visual representation” of children became a Class B felony and a registrable offense. 

States are making these decisions despite President Donald Trump’s executive order in December that aims to preempt states from enacting rules governing AI. The move is a departure from the typical federalist structure of American government, where power is divided between the central government and states, and is likely to see legal challenges.

Maine State Police Lt. Jason Richards, who has investigated child sexual abuse and exploitation for 21 years, has seen the ways in which children can be victimized increase exponentially during that time, he told the Judiciary Committee in November. 

AI has been used to exploit children “in ways never seen before,” Richards said, and when it comes to scrubbing explicit images from the internet, “you can’t really put the genie back in the bottle.”

In 2024, the National Center for Missing and Exploited Children received 67,000 reports of AI-generated child sexual abuse material and 485,000 reports by the first half of 2025, amounting to a 624% increase.

Elon Musk’s artificial intelligence chatbot Grok on X, formerly Twitter, has recently faced global backlash for generating sexualized deepfakes of adults and children. 

However, U.S. officials are embracing it amid the controversy. On Monday, Defense Secretary Pete Hegseth announced Grok will soon operate inside Pentagon networks. 

The bill in Maine was developed in collaboration with the Department of Public Safety. The Maine Chiefs of Police Association, Maine Prosecutors’ Association, Maine Coalition Against Sexual Assault and Maine Office of the Attorney General have publicly supported it.  

However, the Maine Association of Criminal Defense Lawyers is opposed, testifying that the bill is “well-intentioned but dangerously overbroad.”

Specifically, the bill redefines “child sexual abuse material” to include AI-created images that “appear to depict” a minor. The association argues that broad definition risks violating First Amendment protections because it could apply to non-obscene or artistic works that do not exploit an actual child. 

The association cited a 2002 U.S. Supreme Court ruling that struck down portions of a federal child pornography law, deciding that the graphic manipulation of images, even if it makes it appear that children are engaging in sex, is legal and protected speech.