This proposed California ballot initiative aims to protect children who use AI chatbots

A leading child safety advocacy group has teamed with OpenAI to push for a statewide ballot initiative, which they say would be the most comprehensive artificial intelligence safety measure for children in the country.

If the measure is placed on the ballot in November, and if California voters OK it, the Parents & Kids Safe AI Act would require companies to adopt a set of requirements aimed at protecting minors from potentially harmful effects that have been associated with AI use and would authorize the state attorney general to enforce those rules.

Related: Ban social media for kids? This Long Beach lawmaker says Australia is on to something

“At this pivotal moment for AI, we can’t make the same mistake we did with social media, when companies used our kids as guinea pigs and helped fuel a youth mental health crisis in the U.S. and around the world,” Common Sense Media founder and CEO James Steyer said in a statement.

“Kids and teens need AI guardrails now,” he added. “That’s why we will pursue every avenue, from the legislature to the ballot.”

Public interest in kids’ online safety, especially when it comes to AI, has grown over the years, prompted in part by incidents like the death of Adam Raine, a 16-year-old from Orange County whose parents sued ChatGPT maker OpenAI, alleging their son was coached by its chatbot before dying by suicide last year.

Now, the Parents & Kids Safe AI Act would require AI companies to determine if someone using their platform is a minor by using age assurance technology and keep children and teens from becoming emotionally dependent on, or engaging in simulated romantic relationships with, AI systems or having them think they’re talking to a human being.

It also would require safeguards to keep minors from seeing AI-generated content that promotes self-harm, eating disorders, violence and sexually explicit acts; ban advertising that targets minors; prohibit companies from selling data about a minor without parental consent; and give parents control to monitor or limit their children’s use of AI and receive alerts if their children exhibit signs of self-harm.

Furthermore, it would mandate that AI companies undergo independent safety audits and conduct annual child safety risk assessments, and allow the attorney general to investigate and impose financial penalties on companies.

“We look at this as societal seatbelts for the AI era,” Steyer said during a call with reporters this month.

Before launching Common Sense Media, Steyer founded Children Now, a national advocacy and media organization, and was chair and CEO of JP Kids, a leading educational children’s media company. He has also taught reading and math in urban schools and is considered a longtime advocate for children.

During that same press call, OpenAI’s chief global affairs officer, Chris Lehane, said AI is an empowerment tool that helps people solve complex problems. It’s also critical, he said, that parents are empowered in terms of controlling how their children use the technology.

He said OpenAI’s hope is that the safeguards outlined in the proposed ballot initiative won’t just be adopted in California, but that they will serve as a model for other states and even the federal government.

“At the end of the day, I think what this really does speak to is an effort to really stand up for the common good,” Lehane said.

“We are at the front end of the intelligence age. Let’s try to get this right,” he added.

The Parents & Kids Safe AI Act represents the merger of two previous ballot initiatives introduced last year — one by Common Sense Media, the other by OpenAI. Representatives for the two organizations said they decided to work together on a single proposal rather than confuse voters with competing measures.

Common Sense Media’s original ballot initiative was similar to a bill that Gov. Gavin Newsom vetoed last year.

Not long after, OpenAI introduced its own ballot measure, largely reflecting a bill that Newsom had already signed into law last fall. That bill requires companies to inform users that a companion chatbot is AI-generated if a reasonable person could mistake it for actual human interaction and requires them to adopt protocols to prevent chatbots from producing content promoting suicide or self-harm.

In merging the two measures, Common Sense Media dropped two of its previous provisions. One would have prohibited students from using smartphones during the school day. Another provision called for schools to teach AI literacy and safety to help students understand how to use AI responsibly and ethically.

Supporters of the initiative have until June 24 to gather the hundreds of thousands of voter signatures they need to get the proposal onto the November ballot.