OpenAI endorses California youth AI safety bill SB 1119

OpenAI endorsed California Senate Bill 1119 on August 31 and urged Governor Gavin Newsom to sign the companion-chatbot child-safety measure, which state lawmakers gave final approval the same day.

The bill would apply to companies that make companion chatbots — AI systems built to hold sustained, human-like conversations with a user rather than answer one-off queries — available to people in California. Under the text amended in the Assembly on August 28, its main child-protection provisions would become operative on July 1, 2027 if the measure is signed.

OpenAI set out its position in a published statement and in a letter to Newsom dated August 28. “OpenAI strongly supports Senate Bill 1119 and the thoughtful, meaningful safeguards it establishes,” the letter said. Ann O’Leary, the company’s vice president of global policy, said in the statement: “We encourage Governor Newsom to sign the bill into law.”

OpenAI described the measure as requiring companies to determine a user’s age, complete safety-risk work before release, submit to independent audits, protect minors from specified harmful interactions, provide parental tools, connect users to crisis support, limit advertising and protect young users’ personal information. That is the company’s characterization of a pending bill, not a description of current law.

SB 1119 has not been enacted. California’s official bill history records that the Assembly passed the measure and the Senate concurred in the Assembly amendments on August 31, after which it was ordered to engrossing and enrolling; the record shows no action by the governor as of September 2. An August 31 announcement from the office of Senator Steve Padilla, who introduced the bill in February, said it passed the Senate 39-0 and the Assembly 64-4 and was headed to the governor.

The bill text defines a child as a person under 18. A covered operator would have to determine a user’s age through specified statutory mechanisms or apply the designated child protections to all users. Before making a new or substantially modified covered companion chatbot available in California, an operator would have to carry out and document a risk assessment and the mitigations applied.

The text would also restrict cross-context behavioral advertising to children, targeted advertising based on a child’s personal information in conversational chats, the sale of child users’ personal information and certain other uses or sharing of it, and dark patterns affecting safety features and controls.

Audits would follow a fixed schedule. Subject to a temporary exemption through 2031 for operators with less than $500 million in prior-year gross revenue, an operator would need an initial child-safety audit by January 1, 2029, or before it first makes a companion chatbot publicly available, whichever is later, then independent audits every two years and in a specified higher-risk case involving a substantial modification.