OpenAI joins child safety group for California chatbot ballot initiative

Child safety advocate Common Sense Media and ChatGPT creator OpenAI teamed up today to advance a ballot measure that would amend the California Constitution to protect children from companion chatbots online.

The two previously planned to put competing initiatives before voters, each stipulating that the one with the most yes votes would win. The OpenAI proposal largely mirrored existing legislation, while the Common Sense measure included new prohibitions on what AI systems children can access.

The combined measure is known as Safe Artificial Intelligence for Parents and Children Act. This would, among other things:

• Requiring chatbot developers to use technology to estimate a user’s age range and apply filters and security settings for people estimated to be under 18
• Require AI systems to undergo independent audits to identify child safety risks and report them to the California Attorney General
• Prohibit child-targeted advertising and the sale or sharing of children’s data without parental consent
• Stop manipulation through emotional dependency by preventing AI systems from promoting isolation from family or friends, faking romantic relationships with children, or claiming to be sentient

A spokesman for Common Sense said the measure was filed Thursday afternoon. that’s it not yet visible on the Attorney General’s website but you can read a copy obtained by CalMatters here. like described in a press releasethe combined measure removes the ban on smartphones for students in K-12 schools in California and the ban on minors using chatbots capable of engaging in erotic or sexually explicit conversations that were part of Common Sense Media’s original initiative.

The initiative needs 546,651 signatures to go before voters in November. California Secretary of State Shirley Weber has until June 25 to determine whether she meets that threshold or qualifies for the ballot.

Common Sense introduced its initial ballot initiative, the California Artificial Intelligence Safety Act, last fall, shortly after Gov. Gavin Newsom vetoed a bill authored by the nonprofit that contained similar provisions.

In response, in December 2025, OpenAI proposed a competing ballot measure that reflected a bill that Newsom signed into law last Octoberrequiring providers of companion chatbots to implement a suicidal ideation protocol and inform people every three hours that they are talking to an AI. Critics called the move manipulative and designed to thwart stronger protections for children.

A study by Common Sense Media found that seven out of 10 teens have used companion chatbots and that the technology is too dangerous to be used by minors. In promoting its initial ballot initiative, the group warned that without action, the technology could lead to more harm and addiction for young people. In one a well-publicized casethe parents of California teenager Adam Rein sued OpenAI, claiming Rein was coached by OpenAI’s ChatGPT to kill himself.

OpenAI’s willingness to compromise marks a contrast to the way tech companies have come together to navigate policy battles in 2020. Big gig economy players like DoorDash, Instacart, Lyft and Uber spent $200 million that year funding a successful ballot initiative regulating concert operations, Proposition 22. This effectively exempts them from a state law that would require companies to provide full employment benefits to their drivers.

Sen. Steve Padilla, the Chula Vista Democrat who introduced the chatbot bill signed by Newsom, called the combined ballot measure a significant breakthrough. But he added that he thinks the issue should be addressed by lawmakers and the governor, instead of directly by voters. Because the ballot initiative would amend the state constitution, Padilla said it would “create an unnecessarily high bar for revising and updating this law in the future. Additionally, the legislative hearings will provide the broader public with an opportunity to comment and provide input on this important issue.”

In recent weeks, Padilla proposed a bill with a four-year moratorium on the sale of toys with accompanying chatbots inside. OpenAI has signed a partnership with Barbie maker Mattel, but does not produce any products yet.

OpenAI’s battle at the polls in California isn’t limited to children’s online safety issues. One proposed ballot initiative would give a state commission the power to slow or stop the development of an AI model if commission members suspect a catastrophic risk of harm to Californians. Two other proposals target corporate conversions from nonprofits to for-profits, as OpenAI plans. The initiatives force non-profit organizations that restructure in such a way to devote all their assets to the public benefit of humanity. To achieve this goal, the initiatives will create a commission that has the power to rule out AI models and that hosts competitions that invite the public to suggest ways AI can help humanity. Under one of the initiatives, the commission would also have the power to revoke the conversion of non-profit organizations.

OpenAI was founded about a decade ago with a charter stating that its purpose was to benefit humanity. His plans to transform into a public benefit corporation led to heavy criticism from non-profit organizations and review by the California and Delaware attorneys general. Both countries eventually reached agreements with OpenAI to allow restructuring after the company agreed to put approximately 25% of its assets into a non-profit organization.