California Governor Gavin Newsom has vetoed a controversial AI safety bill that aimed to establish groundbreaking regulations for large-scale artificial intelligence models. The decision, announced on Sunday, September 29, 2024, marks a significant moment in the ongoing debate surrounding AI governance and highlights the complex challenges faced by policymakers in crafting effective regulations for this rapidly evolving technology.
The Vetoed Bill: SB 1047
The vetoed legislation, known as SB 1047, was designed to address growing concerns about the potential risks associated with powerful AI systems. It would have required companies developing large AI models to conduct rigorous safety testing and publicly disclose their findings. The bill specifically targeted AI systems that require more than $100 million to build, a threshold that, while not currently met by any existing models, could become relevant within the next year due to the massive investment scale-up within the industry.
Proponents of the bill, including tech luminaries like Elon Musk and AI company Anthropic, argued that it would have introduced much-needed transparency and accountability to the development of large-scale AI models. They emphasized the importance of understanding how these models behave and why, given the potential societal impact of AI technologies.
Governor Newsom's Rationale
In his statement explaining the veto, Governor Newsom acknowledged the bill's good intentions but expressed concerns about its approach. He argued that the legislation did not adequately consider the specific contexts in which AI systems are deployed, such as high-risk environments or critical decision-making processes.
"While well-intentioned, SB 1047 does not take into account whether an AI system is deployed in high-risk environments, involves critical decision-making or the use of sensitive data," Newsom stated. "Instead, the bill applies stringent standards to even the most basic functions — so long as a large system deploys it. I do not believe this is the best approach to protecting the public from real threats posed by the technology."
Industry Opposition and Concerns
The tech industry, including both startups and tech giants, voiced strong opposition to the bill, arguing that it could drive AI companies out of California and hinder innovation. Critics, including former U.S. House Speaker Nancy Pelosi, went as far as to claim that the bill would "kill California tech" and stifle innovation.
Industry representatives expressed concerns that the proposed regulations would discourage AI developers from investing in large models or sharing open-source software. This pushback highlights the delicate balance policymakers must strike between ensuring public safety and fostering technological advancement.
The Global Context of AI Regulation
California's AI safety bill was seen by many as a potential trailblazer for AI regulation in the United States. The European Union has already taken steps to regulate AI, and California's proposal, while not as comprehensive, was viewed as a significant first step in establishing guardrails for the rapidly growing technology.
The veto of SB 1047 raises questions about how the United States will approach AI regulation at both the state and federal levels. It also underscores the challenges of crafting legislation that can keep pace with the rapid advancements in AI technology.
Newsom's Alternative Approach
Instead of signing the bill into law, Governor Newsom announced a partnership with several industry experts, including AI pioneer Fei-Fei Li, to develop alternative guardrails around powerful AI models. This move suggests a preference for a more collaborative approach between government and industry in addressing AI safety concerns.
Newsom emphasized the importance of establishing effective regulations based on "an empirical, science-driven trajectory analysis". He also directed state agencies to expand their assessment of risks associated with potential catastrophic incidents linked to AI applications.
The Debate Over AI Safety Measures
The veto of SB 1047 has reignited the debate over how best to ensure AI safety while promoting innovation. Supporters of the bill argue that mandatory testing and disclosure requirements are necessary to protect the public from potential AI risks, including job loss, misinformation, privacy invasions, and automation bias.
Daniel Kokotajlo, a former OpenAI researcher who resigned over concerns about the company's approach to AI risks, highlighted the unprecedented power that private companies wield through large AI models. "This is a crazy amount of power to have any private company control unaccountably, and it's also incredibly risky," he stated.
The Role of Voluntary Commitments
In the absence of formal regulations, some industry leaders have taken voluntary steps to address AI safety concerns. Last year, a number of leading AI companies agreed to follow safeguards set by the White House, such as testing and sharing information about their models.
However, critics argue that voluntary commitments are insufficient to address the potential risks posed by advanced AI systems. They contend that formal regulations are necessary to ensure consistent safety standards across the industry and to hold companies accountable for the impacts of their AI technologies.
The Future of AI Regulation in California
While the veto of SB 1047 represents a setback for proponents of stringent AI regulation, it does not mark the end of efforts to govern AI development in California. Two other sweeping AI proposals, which faced similar opposition from the tech industry, failed to advance past legislative deadlines last month. These bills would have required AI developers to label AI-generated content and banned discrimination from AI tools used in employment decisions.
The ongoing debate surrounding AI regulation in California reflects broader national and global discussions about how to harness the benefits of AI while mitigating its potential risks. As AI technologies continue to advance and permeate various aspects of society, policymakers will likely face increasing pressure to develop effective regulatory frameworks.
Implications for the AI Industry
Governor Newsom's veto of SB 1047 is seen as a victory for big tech companies and AI developers, many of whom lobbied extensively against the bill. The decision may provide temporary relief for companies concerned about regulatory burdens, but it also leaves open questions about how the industry will address growing public concerns about AI safety and ethics.
The challenge moving forward will be to find a balance between fostering innovation and ensuring responsible AI development. This may involve more targeted regulations that focus on specific high-risk applications of AI, rather than broad mandates that apply to all large-scale models.
The veto of California's AI safety bill underscores the complex challenges involved in regulating emerging technologies. As AI continues to evolve and impact various aspects of society, the debate over how best to govern its development is likely to intensify.
Governor Newsom's decision to pursue a collaborative approach with industry experts may offer a path forward, but it remains to be seen whether this strategy will be sufficient to address the myriad concerns surrounding AI safety and ethics. As California continues to grapple with these issues, its decisions will likely have far-reaching implications for AI regulation across the United States and beyond.
