Colorado moves to ban sexually exploitative AI images, videos

As artificial intelligence expands its scope to more aspects of people’s lives, Colorado legislators want to ensure that state policies on online harassment and child pornography remain at a pace.

The story was produced as part of the Colorado State Capitol Alliance. It first appeared on cpr.org.

“In a modern world where our identity and us exist far beyond our bodies and are vulnerable to attacks on media, social media and text messaging… we must adapt to protect people in these spaces.”

She was there to testify in support of Senate Bill 288. The law, sponsored by Denver’s Democrat majority leader Robert Rodriguez, expands existing laws regarding the posting of intimate images containing material created by AI.

The issue exploded into public conversation last winter when Megastar Taylor Swift’s deepfake came out online. But the creation of false sexual pictures and videos has torn the lives of lesser-known people, including Florida mayors, Texas teachers and high school students across the country.

“Let’s be clear: these images are not harmless. They are not virtual, they are violations,” testified Brownstein, head of the Denver Children’s Advocacy Center.

Supporters say Colorado is late to tackle the issue. The problem of obscene deepfakes is relatively new, but states move quickly to criminalize practices when they are being used to hurt someone. The Colorado bill also makes it illegal for 38 states to use AI to create child pornography.

That provision raises concerns for both the ACLU in Colorado and criminal defense attorneys. They question the ethics of prosecuting people by owning or distributing material that is fully created by computers.

“When you talk about people under the age of 18, how do you age when it’s not connected to a real person?” asked lawmakers along with James Calbach and the state’s office of public advocates. “How do you draw a line around it? If it’s criminalized, then should it be criminalized in the same way as what portrays real humans as harm?”

Bill supporters rebutted that AI is trained with real images with real victims. And they have become increasingly difficult to separate deepfakes from the actual image of abuse as technology improves, making it more important to treat the same under the law.

Controversy over abusive deepfakes arises as Colorado legislators generally try to involve emerging technologies and their potential impacts in people’s lives. Rodriguez works behind the scenes under even more drastic laws. It’s the first effort to put a guardrail on how businesses use AI to make decisions that impact people’s lives.

The bill has not yet been revealed.

In the meantime, Rodriguez’s exploitative content bill unanimously passed its first committee, awaiting a Senate-wide vote.