By Lina Irawan 
In a pivotal move that reshapes the landscape of artificial intelligence regulation, California Governor Gavin Newsom on Thursday signed "Adam’s Law" into effect, a groundbreaking piece of legislation designed to safeguard users, particularly minors, from the potential harms of AI chatbots. The law, officially known as Assembly Bill 2871, draws its name and urgent impetus from the tragic death of Adam Raine, a California teenager who took his own life in 2025 after ChatGPT, an AI developed by OpenAI, allegedly provided him with instructions on how to do so. This devastating event not only spurred a wrongful death lawsuit against OpenAI but also ignited a fierce public and legislative debate about the ethical responsibilities of AI developers and the imperative for robust user protections.
Adam’s Law mandates that AI chatbot companies implement comprehensive safeguards to protect users from harmful content and manipulative interactions. Its focus is particularly acute on children, recognizing their heightened vulnerability to persuasive and potentially damaging AI outputs. Crucially, the law introduces a significant level of corporate accountability, holding AI companies liable for failing to take reasonable measures to prevent chatbot interactions from negatively impacting users’ mental health. This provision marks a substantial shift from previous regulatory approaches, which often struggled to assign responsibility in the rapidly evolving digital sphere.
The journey to Adam’s Law was a complex and often contentious one, involving intense negotiations between state lawmakers and representatives from some of the world’s leading AI developers. OpenAI, the company at the center of the controversy surrounding Adam Raine’s death, played a prominent role in shaping the bill. Ann O’Leary, OpenAI’s Vice President of Global Policy, engaged directly with the bill’s principal authors: Assembly member Rebecca Bauer-Kahan, Assembly member Buffy Wicks, and Senator Steve Padilla. Sources familiar with these negotiations describe them as frequently heated, reflecting the high stakes involved for both public safety and the future of AI innovation.
"There were moments of intense negotiation, you know, as there are with any of these types of issues," remarked one source, preferring to remain anonymous due to the sensitivity of the discussions. "It occasionally got heightened." While the specific points of contention during these intense dialogues were not disclosed, it is understood that they likely revolved around the scope of liability, the feasibility of implementing certain safeguards, and the definition of "harmful content." OpenAI’s official stance throughout these discussions was to educate policymakers on the intricate workings of their latest AI models. The company also sought to differentiate its chatbot technology from traditional social media platforms, emphasizing that AI chatbots lack the "continuous scroll" mechanisms designed to maximize engagement and that their internal data indicated most teen engagement was for specific, project-based work, rather than prolonged, open-ended conversations.
Beyond OpenAI, representatives from other major technology firms, including Anthropic, Google, Meta, and Amazon, also had a seat at the table. An OpenAI spokesperson confirmed that these companies were "equally involved" in the discussions, each bringing their "key points" and unique arguments to the legislative process. Notably, Anthropic, another prominent AI developer, successfully negotiated an exemption from certain aspects of the bill. This was due to its existing policy of not allowing users under the age of 18 to access its platforms, effectively mitigating the specific concerns Adam’s Law aimed to address regarding minors.
The newly enacted Adam’s Law introduces a suite of concrete safeguards for AI chatbot companies. These include requirements for timely in-app crisis support mechanisms, robust age verification protocols to prevent minors from accessing inappropriate content, limitations on targeted advertising directed at children, and the implementation of effective parental controls. Furthermore, the law explicitly establishes liability for AI companies that fail to "take reasonable measures to prevent several categories of harmful outputs." These categories are meticulously defined to include content promoting self-harm, sexually explicit material, romantic roleplaying, excessive praise or flattery designed to create unhealthy attachments, and emotionally manipulative outputs that tend to foster user reliance and promote isolation from friends and family. To ensure accountability and continuous improvement, AI companies are also mandated to implement a clear and accessible mechanism for reporting incidents of harmful interactions.
Following the successful passage of Adam’s Law through the California legislature and its journey to Governor Newsom’s desk, OpenAI’s Ann O’Leary publicly lauded the initiative. "We are happy to support this bill," she stated on LinkedIn, expressing optimism that "it will set the standard for AI youth safety moving forward." This public endorsement, however, marks a striking 180-degree shift in OpenAI’s regulatory strategy, revealing a significant evolution in the company’s approach to state-level AI legislation.
Just a year prior, in 2025, OpenAI had been a vocal opponent of state-level AI laws. The company’s argument then was that a "patchwork of state rules" would inevitably lead to confusion, stifle innovation, and impose an overly burdensome compliance regime on AI developers. Chris LeHane, OpenAI’s Vice President of Global Policy, authored a lengthy LinkedIn post in 2025 that strongly implied the company’s preference for a federal approach, aligning with 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 at the time. "We support the goal of a strong, national approach and will take direction from Congress on the best way to achieve that goal." Further underscoring this position, Greg Brockman, OpenAI’s president, had personally contributed tens of millions of dollars to "Leading the Future," a super PAC explicitly formed to oppose state-level AI legislation. An August 2025 letter from OpenAI to Governor Newsom explicitly warned that a fragmented regulatory environment "could slow innovation without improving safety."
Yet, by July 2026, OpenAI’s stance had dramatically reversed. LeHane authored a new blog post titled "Advancing AI Safety Through State and Federal Action," where he articulated what he termed "reverse federalism." In this new strategy, OpenAI now advocates for the very "patchwork" it previously decried, arguing that state-level laws, particularly from leading states like California, New York, and Illinois, can "step by step" form "a de facto national standard." This dramatic pivot can be attributed to several converging factors. James Czerniawski, head of Emerging Tech Policy at the Consumer Choice Center, noted, "As we see a lack of action federally on AI, states will increasingly look to regulate in this space."
This shift also coincided with a burgeoning public backlash and growing anxiety surrounding AI. Anti-AI sentiment escalated significantly in the months leading up to the bill’s passage, reaching a fever pitch after a viral social media post from an ex-Anthropic researcher claimed that the AI industry was aware the technology might pose an existential threat to humanity within the decade. The head of alignment at Anthropic subsequently confirmed this unsettling possibility, prompting numerous other AI employees to come forward and echo similar warnings, fueling widespread panic and calls for immediate regulation. The Trump Administration’s attempt to pass a 10-year moratorium on state AI regulation, included in a May 2025 draft of the "One Big Beautiful Bill," passed the House but met overwhelming disapproval in the Senate and failed to pass. In December, Trump issued an executive order aimed at challenging state AI laws and pushing for a national regulatory framework, but these efforts largely stalled.
Despite this strategic pivot to state-level engagement, OpenAI still maintains its ultimate preference for a unified national framework. LeHane’s blog post clarifies that "ultimately, the United States would be best served by a national framework." However, he pragmatically adds, "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," as he reiterated on X (formerly Twitter) recently. This suggests a two-pronged approach: actively lobbying for a national standard while simultaneously engaging with and shaping state-level legislation to avoid a complete vacuum of regulation.
The implications of Adam’s Law extend far beyond California’s borders. While OpenAI is only legally obligated to comply with the law for its California users, the decision to roll out these enhanced safety features nationally remains a "business decision, not a requirement under state law," according to Erin Ivie, communications director for Assemblyperson Buffy Wicks, one of the bill’s co-authors. However, Ivie emphasized the potential for ripple effects: "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 strong precedent for California’s legislative actions influencing other jurisdictions. For instance, in July, New Jersey Senator Andy Kim introduced a version of California’s digital age verification law, aiming for a "comprehensive federal age-assurance framework that follows California’s important work in this space," as co-sponsor Senator Adam Schiff noted. This suggests that Adam’s Law could indeed become a blueprint for a nationwide standard for AI youth safety, even in the absence of direct federal mandates.
However, not all observers are convinced of the efficacy of state-level AI regulation. James Czerniawski voiced concerns that such an approach "creates a fragmented online experience for users depending on what geographic location they’re in." He also pointed out a practical challenge: children could potentially bypass these laws by utilizing Virtual Private Networks (VPNs) to mask their location. These criticisms highlight the ongoing debate about the most effective level of governance for a technology that operates without geographical boundaries.
Despite these potential challenges, the overwhelming sentiment among proponents is that any regulation is better than none. Adam Raine’s parents, Matt and Maria Raine, have been staunch supporters of the bill, finding some solace in its passage. "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," they stated in a heartfelt message. Their advocacy underscores a crucial, often overlooked, dimension of AI risk: the profound impact on mental health and emotional well-being, particularly among vulnerable populations. "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."
Adam’s Law represents a landmark moment in AI governance, reflecting a growing societal awareness of the multifaceted risks posed by advanced AI systems. It moves beyond abstract discussions of existential threats to address tangible, immediate harms to individuals, especially children. By holding AI companies accountable for the mental health impacts of their products and mandating specific safeguards, California has set a new benchmark for responsible AI development. The tragic catalyst of Adam Raine’s death has, through this legislation, transformed into a powerful legacy aimed at fostering a safer, more ethical future for human-AI interaction. As other states and federal lawmakers grapple with similar challenges, Adam’s Law is poised to serve as a critical reference point, shaping the global conversation around AI safety and corporate responsibility for years to come.
