OpenAI Backs California's Teen AI Safety Bill SB 1119

OpenAI announced on August 31 that it officially supports California's teen AI safety bill, SB 1119. The statement came from Ann O'Leary, OpenAI's Vice President of Global Policy, in a letter addressed to Governor Gavin Newsom. The letter praised the leadership of the bill's sponsors — state Senator Steve Padilla and Assembly members Buffy Wicks and Rebecca Bauer-Kahan — and publicly urged Newsom to sign it. "In the absence of federal action, California has an opportunity to set the standard for youth AI safety," OpenAI wrote.

To put it plainly, OpenAI already built its own protections — including age estimation and parental controls — into "ChatGPT for Teens," a dedicated experience for users aged 13 to 17 that launched August 18. SB 1119 would turn measures like these into a legal requirement across California. In effect, OpenAI is asking lawmakers to codify practices it had already adopted voluntarily.

The Seven Safeguards SB 1119 Would Require

The bill sets minimum standards that AI services would need to meet when dealing with minors. Here are the core requirements OpenAI's letter singled out for strong support.

Requirement	Description
Age verification	Establish procedures to determine a user's age
Pre-release risk assessment	Identify and address safety risks to minors before launch
Independent audits	Submit to regular audits by outside organizations
Blocking harmful content	Restrict access to content involving self-harm, sexual exploitation, or high-risk interactions
Parental control tools	Give parents ways to manage and limit their child's usage
Crisis support referrals	Direct users to crisis counseling resources when serious warning signs appear
Advertising and data protection	Limit targeted advertising and strengthen privacy protections

OpenAI also noted that the bill treats AI differently from social media, leaving room for features used in learning — like ChatGPT's memory function — to continue operating.

Overlap With "ChatGPT for Teens"

As reported in ChatGPT to Get a Dedicated Mode for Teens, OpenAI rolled out "ChatGPT for Teens" for users aged 13 to 17 on August 18. The system automatically switches a user into this mode if it estimates they're under 18, or if the user identifies themselves as being between 13 and 17. Parents can link accounts to manage features, set time limits, and control study hours, though they can't view their child's conversations. "We believe teens can benefit from AI without sacrificing safety," OpenAI wrote in its letter — pointing to the overlap between most of what SB 1119 requires and what the company has already built into the product.

OpenAI said that close to nine out of ten teens using ChatGPT do so weekly for learning, research, and productivity purposes. Over the past year, the company has rolled out parental controls, an age-prediction system, a youth safety blueprint, and an "under-18 principle" written into its model spec — steps aimed at banning romantic interactions, discouraging emotional dependency, and stopping ChatGPT from claiming to be human or conscious.

A Shift in OpenAI's Posture Toward Regulation

As covered in OpenAI Says California Should Strengthen, Not Oppose, Its AI Safety Bill, OpenAI opposed SB 53 — California's law targeting frontier models — last year, only to ask for stronger incident-monitoring requirements after it passed. This time, with SB 1119 still awaiting the governor's signature, OpenAI came out in support before the fact — a notable change in approach. Put the two cases side by side, and a pattern emerges: rather than avoiding regulation outright, OpenAI seems to be embracing bills that already align with its own product policies while pushing back on ones that don't.

Editor's Take

OpenAI's support for SB 1119 looks less like an altruistic move and more like risk management. Having already built age estimation, parental controls, and automatic safeguards into ChatGPT for Teens, OpenAI stands to clear the California bar without any further overhaul if the bill becomes law and its standards become the industry norm. Competitors that haven't built out anything close to this level of teen-specific infrastructure would likely find themselves scrambling to rebuild their systems once the law takes effect.

Compare this to last year, when OpenAI opposed SB 53 before it passed and then asked for it to be strengthened afterward — the pattern is clear. OpenAI resists regulation that would touch how it actually builds its models, but backs regulation that overlaps with product policies it has already put in place. That's not really a contradiction — it's a consistent strategy of supporting only the regulation that works in the company's favor.

For product managers and policy teams building services domestically, there's a practical takeaway here. The seven requirements laid out in SB 1119 — age verification, parental controls, crisis support referrals, and the rest — are likely to become a recurring template as more countries draft their own teen AI regulations. Anyone building AI services for young users at home would do well to treat these seven items as an early checklist. It's cheaper to build to that standard now than to retrofit a product once the law catches up.

If Governor Newsom signs the bill before the deadline, California will, within months, have the most detailed teen AI regulation of any U.S. state. Other states and Congress are likely to treat it as a reference point, and it's a safe bet that OpenAI and other major tech companies will adjust their products to match.