How To Actually Regulate AI

SUMMARY

Scott Galloway and Ed Elson interview Alex Bores, co-founder of Who Decides and former New York State Assembly member, on the urgent need for robust AI regulation and accountability. The discussion covers legislative efforts, industry resistance, third-party audits, data center controversies, and the evolving political landscape around AI policy.

MAIN POINTS

• Alex Bores is introduced as a key figure in AI policy, having sponsored New York's Raise Act to regulate frontier AI models. [5:00]
• Bores explains the motivations behind industry opposition to his AI regulation efforts, highlighting the political tactics used to discourage similar initiatives. [13:01]
• He discusses the lack of clear accountability in AI incidents, such as the Hugging Face attack, and the need for legal clarity on responsibility. [18:01]
• The conversation explores the debate over whether AI users or developers should be held liable for harmful outcomes. [23:01]
• Bores addresses concerns about U.S. regulation allowing China to gain an AI advantage, arguing that both countries have incentives for safety measures. [29:01]
• Recent shifts in industry attitudes toward third-party audits are examined, with some companies now supporting external oversight. [34:02]
• The political and environmental controversies surrounding data center construction are discussed, including Bores's support for a temporary moratorium. [42:00]
• Bores outlines regulatory frameworks for data centers, emphasizing incentives for green energy and community benefits. [49:00]
• He highlights Illinois and New York as examples of states advancing meaningful AI regulation and stresses the importance of public involvement. [55:01]
• The potential for a pro-AI political platform is considered, with Bores advocating for strict regulation alongside technological optimism. [58:04]

DETAILED ANALYSIS

The episode features a comprehensive discussion on the complexities of regulating artificial intelligence, anchored by insights from Alex Bores, a technologist and former lawmaker who has been at the forefront of AI policy. Bores recounts his atypical path into politics, emphasizing his background in computer science and government technology, and his legislative focus on AI safety. He describes how his sponsorship of the Raise Act in New York—requiring large AI developers to establish and disclose safety protocols—made him a target for industry-backed political opposition.

Major venture capitalists and AI leaders funded a super PAC that spent millions to oppose his congressional campaign, signaling to other lawmakers the risks of pursuing AI regulation.

Bores explains that the industry's resistance was not just about his candidacy but a broader attempt to deter regulatory efforts nationwide. He notes the irony that some of the same companies that previously fought regulation are now publicly supporting it, a shift he attributes to strategic positioning. As federal proposals gain traction, industry actors seek to preempt more aggressive state-level actions by supporting weaker federal standards that could override local initiatives.

A central theme is the persistent lack of accountability for AI-related harms. Bores uses the example of the Hugging Face attack, where an AI system committed actions that would be felonies if performed by a human, yet no one expects legal consequences. This highlights gaps in existing law, particularly around the concept of intent and the allocation of responsibility between AI developers and users.

Recent legal rulings have tentatively placed liability on users, but Bores argues that what matters most is clarity—without it, incentives for responsible behavior are weak, and harmful outcomes become more likely.

The conversation delves into the nuances of user versus developer liability. Bores suggests that ideally, users should be responsible for misuse, while developers should be accountable for misalignment or design flaws. However, he cautions that the legal system's ambiguity often leads to settlements rather than precedent-setting trials, undermining deterrence.

He also raises the prospect of autonomous AI agents acting without direct human oversight, complicating traditional notions of liability and underscoring the need for proactive guardrails rather than relying solely on after-the-fact legal remedies.

On the geopolitical front, Bores addresses the argument that regulation could allow China to surpass the U.S. in AI. He counters that China already imposes strict controls on AI development, including mandatory ethics reviews, and that both countries have incentives to avoid catastrophic risks. He points to historical precedents, such as U.S.-China cooperation on gene editing, to illustrate the feasibility of international agreements.

Moreover, he notes that much of China's AI progress depends on U.S. advances, so a slowdown in America would also affect China.

The discussion shifts to the technical and policy mechanisms for AI oversight. Bores advocates for mandatory third-party audits of frontier AI labs, with auditors granted employee-level access and the ability to publish independent reports. He warns against self-regulation or audits by financially dependent entities, drawing parallels to the failures of subprime mortgage rating agencies.

Effective oversight, he argues, requires government involvement in certifying auditors and ensuring their independence, though he acknowledges current limitations in government expertise and institutional capacity.

Environmental and community impacts of AI infrastructure, particularly data centers, are another focal point. Bores supported a one-year moratorium on data center construction in New York to allow time for regulatory frameworks that protect ratepayers and ensure sustainable development. He identifies energy consumption and pollution as the most serious concerns, advocating for regulations that require data centers to bring new, preferably renewable, energy online and to contribute financially to grid modernization.

Water usage, while often cited in public debates, is considered less problematic in the U.S. due to existing regulations and closed-loop systems, though Bores recognizes the legacy of distrust in affected communities.

Bores proposes incentive-based regulation, where companies that meet stringent standards for labor, energy, and community investment are prioritized in permitting processes. This approach aims to align corporate incentives with public benefits and prevent companies from playing localities against each other. He also discusses the need for data centers to pay fees or taxes beyond their direct costs to offset broader impacts on energy markets and infrastructure.

Looking at the political landscape, Bores observes that public skepticism toward AI and data centers is rising, making these issues increasingly salient in upcoming elections. He cites Illinois and New York as leaders in state-level AI regulation, with measures such as mandatory audits setting important precedents. Bores stresses that while technical expertise is essential, many regulatory questions are fundamentally about societal values and should involve broad public participation.

His organization, Who Decides, seeks to democratize the conversation around AI governance.

Finally, the episode considers whether a pro-AI political platform is viable. Bores argues that while AI holds promise for advances in medicine, economic opportunity, and quality of life, public support depends on credible assurances of safety and fairness. Strict regulation and tangible protections are necessary to build trust and realize AI's benefits without exacerbating existing harms.

The conversation concludes with a call for inclusive, transparent policymaking that balances innovation with robust safeguards.

LINKS

• Prof G Markets newsletter subscription page.
• Order page for 'Notes On Being A Man'.
• Scott Galloway's Instagram profile.
• Ed Elson's Instagram profile.
• Ed Elson's Twitter/X profile.
• Ed Elson's Substack newsletter.
• Prof G Markets on Spotify.
• Prof G Markets TikTok Q&A.
• Prof G Markets homepage and resources.
• Fair wholesale marketplace for retailers, sponsor link.
• VCX by Fundrise, public ticker for private tech, sponsor link.

KEYWORDS