OpenAI and Common Sense Media tried to turn a California child safety fight into a ballot measure. The measure failed before voters ever saw it.
California's AI ballot fight no longer ends in November. On August 21, the California Secretary of State's office said the petition for the Parents & Kids Safe AI Act had failed because supporters filed fewer signatures than the state required. That's the hard fact now. If you read the January announcement as OpenAI choosing regulation, you have to read the August failure as something plainer: the company backed a rulemaking route that didn't survive the signature stage.
The proposal began on January 9, when OpenAI and Common Sense Media announced they had merged two competing ballot measures into one. The combined measure would have required AI providers to estimate users' ages, give parents controls over children's use, bar child-targeted advertising, limit the sale or sharing of minors' data without parental consent, and submit to annual independent audits reported to the California attorney general.
Those are not small asks.
According to the Secretary of State's February circulation notice, proponent Thomas W. Hiltachk needed 546,651 registered voter signatures by August 10 for the initiative to become eligible for the ballot. The state later said the petition came up short. So the question has changed. It isn't whether Californians will vote on this OpenAI backed measure in November. They won't, unless some separate legal twist revives a dead petition.
OpenAI Cuts GPT-5.6 Sol API Prices After Holding the Line for Months OpenAI cut API pricing on its flagship GPT-5.6 Sol model by more than 20% starting August 21, 2026, the first price cut to its top-tier model since launch. The move follows earlier discounts to its Terra and Luna models and comes as Anthropic and Chinese labs like DeepSeek and Moonshot AI undercut OpenAI on enterprise API pricing. - OpenAI GPT-5.6 Sol API price cuts announced - when did OpenAI reduce GPT-5.6 Sol pricing
The compromise mattered even before it failed #
The failure doesn't make the January deal irrelevant. It shows you how the biggest AI companies are learning to handle regulation: fight some bills, narrow others, and back the version that gives them the most predictable rules.
OpenAI had already spent serious energy warning against California's state-by-state approach. In 2024, it opposed SB 1047, the broader AI safety bill that Governor Gavin Newsom later vetoed. In August 2025, OpenAI published a letter to Newsom arguing that a patchwork of state rules could slow innovation without improving safety. Newsom signed SB 53 anyway on September 29, 2025, creating what his office called a first-in-the-nation frontier AI transparency law.
Then OpenAI went to the ballot.
In December 2025, the company filed its own initiative tied closely to California's companion chatbot law, SB 243. State Senator Steve Padilla, who authored SB 243, didn't treat that as a friendly gesture. In a December statement, Padilla called OpenAI's move Big Tech's latest attempt to limit commonsense regulation of dangerous AI chatbots and said it was trying to halt further efforts to protect children and others from the tools.
That attribution checks because Padilla said it in his own office's release. It also fits the politics. A voter-approved statute is harder for lawmakers to change later than an ordinary bill. If you can write the ballot language, you don't just shape one law. You shape the next fight.
Common Sense Media saw the compromise differently. Its January release called the Parents & Kids Safe AI Act the strongest youth AI safety measure in the United States, and founder James P. Steyer said the group welcomed OpenAI's support rather than confusing voters with rival proposals. That's a reasonable public line. Common Sense got OpenAI behind age assurance, parental controls, audits, and attorney general enforcement.
But the merged proposal was narrower than what children's advocates first wanted. CalMatters reported that the combined measure dropped a K-12 smartphone ban and removed language that would have barred minors from using chatbots capable of erotic or sexually explicit conversation. What survived was still meaningful. What disappeared was meaningful too.
IBM Partners With OpenAI To Put GPT-5.6 Inside Its Consulting Business IBM and OpenAI announced a partnership on August 13, 2026, integrating GPT-5.6, Codex and ChatGPT Work into IBM Consulting Advantage. The deal follows IBM's prior alliance with Anthropic and a separate $240 million inference infrastructure deal with Together AI, underscoring IBM's bet on multi-vendor AI integration over building its own frontier... - IBM OpenAI partnership consulting services - enterprise AI implementation for banks
The fight moved back to lawmakers and products #
California still has rules on the books. SB 243 took effect on January 1, 2026, with requirements for companion chatbot operators to disclose when users are interacting with AI and follow self-harm protocols. Minors also get break reminders during long sessions. SB 53 adds transparency duties for large AI developers. These laws matter because the ballot measure is gone.
The product side is moving as well. The Associated Press reported this week that OpenAI has introduced ChatGPT for Teens, a version for users aged 13 to 17 with tighter limits around self-harm, suicide, romantic or sexual content, and other sensitive areas. OpenAI's own parental controls page also says parents can link accounts and set quiet hours, with narrow safety notifications for some urgent situations.
Frankly, that's where the pressure belongs now. A failed ballot measure doesn't make AI chatbot risks disappear. It just removes one carefully negotiated path for handling them.
You don't need to believe every criticism of OpenAI to see the pattern. The company opposed broader California AI safety rules, backed a child safety initiative after filing its own narrower measure, and then watched that initiative fail for lack of signatures. Common Sense Media still has credibility on children's online safety - that part hasn't moved. OpenAI still has the product millions of teenagers use, whatever the ballot measure's fate. And California still has lawmakers who want more.
The next fight will be less tidy than a ballot campaign. It will move through Sacramento hearings, attorney general enforcement, product launches, lawsuits, and whatever families see on their kids' screens. That's less clean for OpenAI. It may be better for everyone else, because ordinary legislation can still be argued over, amended, and strengthened when the facts change.
Also read: Nearly 6,000 CEOs Say AI Hasn't Boosted Productivity, Yet Layoffs Continue • Fortinet Buys Virtue AI Two Months After Meta Poached Its Founders • Stripe's Product Chief Says AI Agents Will Kill the Checkout Page