California’s new ‘Adam’s Law’ on chatbots shows OpenAI’s strategy shift on state AI regulations

The law requires AI chatbot companies to adopt safeguards to protect users—especially children—from harmful content and manipulative interactions, while holding companies liable for failing to take reasonable measures to prevent chatbot interactions from harming users’ mental health.

OpenAI lobbied to shape the bill as part of its latest regulatory strategy to influence state-level bills. Ann O’Leary, OpenAI’s Vice President of Global Policy, worked with its authors, Assembly member Rebecca Bauer-Kahan, Assembly member Buffy Wicks, and Senator Steve Padilla.

Sometimes the conversations got heated, according to sources familiar with the negotiations.

“There were moments of intense negotiation, you know, as there are with any of these types of issues,” said a source familiar with the negotiations. “It occasionally got heightened.”

The source was unable to disclose which points were most contentious. OpenAI said its role in the conversations was to educate policymakers on how the latest AI models work. The company also clarified how it differs from social media, in that there is no continuous scroll, and their data shows most teens engage with the technology to work on specific projects.

Representatives from Anthropic, Google, Meta, and Amazon also had a seat at the table and were “equally involved” in the discussions, an OpenAI spokesperson tells Fortune. Each had their own “key points” and unique arguments. Anthropic, for example, was able to negotiate out of having to abide by the bill because it does not allow users under 18.

Adam’s Law introduces several safeguards for AI chatbot companies. For example, they must have timely in-app crisis support, age verification, limitations on targeted advertising to children, and parental controls. It also introduces liability for AI companies if they fail to “take reasonable measures to prevent several categories of harmful outputs, including self-harm, sexually explicit material, romantic roleplaying, excessive praise or flattery, and emotionally manipulative outputs that tend to foster reliance and promote isolation from friends and family,” according to the announcement. AI companies must also implement a mechanism to report incidents.

After Adam’s Law cleared the California legislature and headed to Newsom’s desk, O’Leary praised the effort. “We are happy to support this bill,” she said on LinkedIn. “We believe that it will set the standard for AI youth safety moving forward.”

A 180-degree change in OpenAI’s regulatory strategy

OpenAI’s interest in shaping state regulations is an abrupt departure from its focus on stopping state-level AI laws just one year ago. At the time, OpenAI was arguing that regulating AI at the state level would sow confusion and create too high a compliance burden on AI companies. Chris LeHane, the company’s vice president of global policy, wrote a lengthy post on LinkedIn in 2025 that strongly suggested the company favored the efforts by some Congressional Republicans and the Trump White House to impose a moratorium on state-level AI regulations.

“Recent proposals like a federal moratorium reflect how seriously Congress is taking this issue,” LeHane wrote. “We support the goal of a strong, national approach and will take direction from Congress on the best way to achieve that goal.” Meanwhile, Greg Brockman, OpenAI’s president, had personally donated tens of millions of dollars to a super PAC, Leading the Future, that opposed state-level AI laws.

In an August 2025 letter to Newsom, OpenAI warned that a “patchwork of state rules…could slow innovation without improving safety.” But now, OpenAI advocates for that exact patchwork, saying it will “step by step” form “a de facto national standard,” according to a July 2026 blog post authored by LeHane.

“As we see a lack of action federally on AI, states will increasingly look to regulate in this space,” James Czerniawski, head of Emerging Tech Policy at the Consumer Choice Center, tells Fortune.

LeHane calls the AI lab’s new approach “reverse federalism,” and names California, New York, and Illinois in its post as examples of states that are on the forefront of AI policy. This shift has accompanied a growing backlash against AI, including data centers. Anti-AI sentiment escalated to panic and anxiety this month after a viral social media post from an ex-Anthropic researcher who claimed the AI industry is aware the technology may kill all humans within the decade. The head of alignment at Anthropic confirmed that is the case, and multiple other AI employees came out of the woodwork to echo the message as well.

The Trump Administration attempted to pass a 10-year moratorium on states passing any AI regulation, including it in a May 2025 draft of the “One Big Beautiful Bill.” It passed in the House but was met with overwhelming disapproval in the Senate and did not pass. In December, Trump issued an executive order aimed at challenging state AI laws and pushing for a national regulatory framework.

OpenAI still supports the national framework—LeHane writes that “ultimately, the United States would be best served by a national framework.” However, he says that “in the absence of one, states can move us there by passing laws that mirror one another.” CEO Sam Altman continues to advocate for a federal framework that “sets consistent safety requirements for frontier AI,” he wrote on X last night.

Chatbot law could be a model for other states

OpenAI must comply with the law for California users only. If they choose to roll out these features nationally that would be “a business decision, not a requirement under state law,” Erin Ivie, communications director for state assemblyperson Buffy Wicks, one of the bill’s co-authors, tells Fortune. “Now that the law has passed, other states, or the federal government, may use our bill as a model and pass their own version.”

There is precedent for California’s laws inspiring other states to adopt similar ones. In July, New Jersey Senator Andy Kim introduced a version of California’s digital age verification law. It’s “a comprehensive federal age-assurance framework that follows California’s important work in this space,” said Senator Adam Schiff, a bill co-sponsor.

However, some are skeptical that state-level AI regulation can be effective. “I think it’s problematic insofar as it creates a fragmented online experience for users depending on what geographic location they’re in,” said Czerniawski. He notes that kids can get around the laws as well by using Virtual Private Networks (VPNs).

Others say any regulation is better than none, and Adam’s parents strongly supported the bill. “We still have not adjusted to life without Adam, but we are pleased that an element of his legacy is to help make AI chatbots safer for minors,” said Matt and Maria Raine. “We believe the risks of unregulated AI companionship rank right up there with other more discussed AI risks, and we are confident Adam’s Law will save lives and prevent other harms.”