Today in AI: California Senate Bill 1119 Backed

OpenAI says it backs California Senate Bill 1119 in an August 31 letter to Governor Gavin Newsom.

Ann O'Leary, the company's VP of Global Policy, framed the bill as an age-appropriate standard in the absence of federal action.

What the bill would require

SB 1119 pairs continued access to useful AI tools with mandatory youth safeguards.

It would require providers to determine a user's age and identify safety risks before launch.

It mandates independent audits and automatic protections for users identified as 13 through 17.

Those protections include blocking self-harm content, sexually exploitative material and other high-risk interactions.

The bill also requires parental guidance tools, crisis-support connections and limits on targeted advertising and data use.

Policymakers drafted it to preserve educational and safety-critical features, including memory.

The gap to watch is how age estimation, audits and enforcement get scoped during implementation.

Inside ChatGPT for Teens safeguards

ChatGPT for Teens automatically places estimated under-18 users and declared 13 to 17 year olds into a distinct experience.

Those protections are baseline, not optional settings users can turn off.

The product adds Quizzes, Learning Visualizations and Study mode built with educators to guide step-by-step problem solving.

Memory carries context across sessions to keep safeguards and learning support consistent over time.

Model rules prohibit romantic engagement, encouragement of emotional dependence and claims of being human or sentient.

Parents get controls over key settings and notifications in high-risk situations.

Sam Altman, OpenAI co-founder and CEO, outlined teen principles last year that this support extends.

Why an AI-specific youth framework matters next

OpenAI argues AI is not social media and needs tailored rules that preserve learning and creation.

That distinction matters for builders who rely on persistent context for tutoring and longer-term projects.

SB 1119 tries to treat continuity features like memory as safety infrastructure, not just personalization.

The tradeoff is compliance overhead for startups that must add age checks, risk reviews and audits.

Nearly nine in ten teens who use ChatGPT already use it weekly for learning, information or productivity, according to OpenAI.

With no federal standard, Newsom's signature decision will set the operational test for other states.

OpenAI says continued study and collaboration will be needed to keep provisions effective as models change.