Law cracking down on AI abuse sails through Congress, White House

Sometimes, an issue takes Congress by storm. Eleanor Gaetan, director of public policy at the National Center on Sexual Exploitation, is thankful that protecting victims of AI-generated pornography is one of those issues.

“If I had two minutes of a Disney classic—like Beauty and the Beast—I can’t put it up on YouTube. It was immediately rejected. It’s taken down because there’s copyright law, and everyone agrees that the copyright protects the intellectual property of Disney. But pictures of me that were naked are not protected by any law. That’s wrong,” Gaetan said.

On Monday afternoon, President Donald Trump signed the Take it Down Act into law. The act makes it a crime to use software, machine learning, artificial intelligence, or other computer-generated technology to create an explicit depiction of a real person without his or her consent. It passed the legislative process with relative ease.

“With the rise of AI image-generation, countless women have been harassed with deep fakes and other explicit images, distributed against their will. This is just so horribly wrong. It’s a very abusive situation. Today we’re making it totally illegal,” Trump said at the White House.

Rep. Jay Obernolte, R-Calif., one of the bill’s co-sponsors, called the act a meaningful first step but said he hopes more work will follow.

“It is one of the first pieces of AI-related legislation,” Obernolte said. “In one way, I am very happy; obviously, as someone who’s been very vocal in expressing my belief that nonconsensual intimate imagery is something we should all be able to agree is not okay. On the other hand, I think it’s very unfortunate this is all that we’ve been able to get done so far.”

The law sets two levels of enforcement: penalties for individuals who create and post the pornographic material and punishment for the companies that enable its propagation.

Individuals creating explicit materials from the image of an adult could face a fine of up to $250,000 and two years behind bars—a sentence that climbs to three years for creating an image involving a minor.

For companies that host the content, the law requires compliance within a two-day window of a complaint. A failure to comply within that 48-hour span would be considered an unfair or deceptive business practice and could result in a $50,000 fine per violation.

Companies making a good-faith effort to remove images within the two-day window would not be held liable.

Some lawmakers who support the act on principle said they couldn’t support those penalties without clearer definitions of the crime. “I mean, they’re talking about AI-generated porn that resembles a person, so how do you decide if it’s actually a real person or someone that looks like a real person? But someone could go to prison if the image looks like somebody? It’s so vague,” said Rep. Thomas Massie, R-Ky., one of the two members in the House of Representatives who voted against the bill.

In particular, Massie believes the two-day requirement puts a massive burden on start-ups who may not have the resources to keep up with potential violations by their user base.

He noted that other members had expressed similar alarm but declined to say which ones. Only one other member, Rep. Eric Burlison, R-Mo., voted against it.

The bill passed by unanimous consent in the Senate.

Responding to Massie’s concern about the two-day window, Obernolte—the bill’s sponsor—believes the cutoff had to be set somewhere to protect the interests of victims.

“Certainly, the courts will work it out,” Obernolte said. “But to anyone who objects to the two-day provision, my question is—well, what is the right amount of time? Because [to] some people that are objecting in their minds, the right amount of time is no amount of time. I think we will have to agree to disagree, but the limitation has to be something.”

Gaetan, director at NCOSE, believes the tech industry will comply with the law by developing means of identifying or flagging problematic content—even before it goes live.

“It is absolutely within the capacity of these companies to remove it before it’s even uploaded,” Gaetan said. “There are extremely successful AI tools to identify imagery that would comply with this law and get rid of it. So, we believe in the incredible success and genius of these companies to be able to solve so many problems, including this one.”

Experts like Clare Morell of the Ethics and Public Policy Center warn that the law can’t ultimately take the place of parental judgment.

“Be really careful who you share photos with, who you text photos to, and be very careful uploading them online. I personally would advise parents to not allow their children under 18 on social media for this reason. But if they are, then they should be really cautioned about being careful knowing that that can, sadly, be used against them,” Morell said.