Skip to main content

AI Regulation: SB 813, Independent Verification, and the Future of Governance

LawfareJuly 29, 202554 min229 views
25 connectionsยท40 entities in this videoโ†’

The Current AI Regulatory Landscape

  • ๐Ÿ’ก The AI community is awaiting the federal government's AI action plan, currently operating in a "weird regulatory void" with numerous state-level AI bills proposed.
  • โš ๏ธ A federal AI moratorium, which would have preempted state legislation, did not pass, leading to a complex patchwork of proposed laws.

Fathom's Novel Approach to AI Governance

  • ๐ŸŽฏ Fathom aims to develop novel solutions for AI governance, recognizing it as potentially the hardest regulatory challenge the world has faced.
  • ๐Ÿง  The approach acknowledges limitations in government expertise and speed, proposing a system that doesn't rely solely on traditional, lengthy legislation.
  • โš–๏ธ Senate Bill 813 (SB 813) in California is presented as a conceptual outline for this new approach, focusing on two key ideas.

Key Components of SB 813

  • ๐Ÿค The first idea is creating a virtuous feedback loop for accountability within a "private regulatory market" or "independent verifier market."
  • โœ… This aims to ensure that private entities setting best practices are held accountable for societal good and demonstrable outcomes.
  • ๐Ÿš€ The second idea is to use "carrots" rather than strict regulations to encourage compliance, focusing initially on tort law as a cross-state framework.

Addressing the Patchwork Problem

  • ๐ŸŒ Startups and entrepreneurs fear a patchwork of state-specific AI regulations, similar to the privacy world, leading to complexity and varied responsibilities.
  • ๐Ÿ—บ๏ธ SB 813 proposes that states could band together to create a cohesive network of subject matter experts, offering a unified approach instead of 50 different solutions.
  • ๐Ÿงฉ This could allow a small startup to gain approval through one private organization recognized by multiple states, simplifying compliance.

Challenges and Opportunities in Verification

  • ๐Ÿ”ฌ Current benchmarking and evaluation organizations often focus on specific risks and may be divorced from the deployer landscape.
  • ๐Ÿ“ˆ There's a concern that a few organizations could dominate standard-setting, potentially leading to a "race to the bottom" or groupthink.
  • ๐Ÿ’ก The proposed framework emphasizes proving good outcomes rather than just checking boxes, allowing for a dynamic and adaptive system.
  • ๐Ÿ› ๏ธ The goal is to create a regulated marketplace that accelerates adoption and ensures alignment with societal good, rather than solely relying on industry-driven standards.

The Role of Government and Philanthropy

  • ๐Ÿ’ฐ Andrew Freedman emphasizes that government's role is to set desired societal outcomes and reward those who achieve them, fostering a public-private partnership.
  • ๐ŸŒŸ Lauren Wagner advocates for significant philanthropic investment to build a dynamic ecosystem of evaluation and verification companies before codifying legislation.
  • โš ๏ธ Both agree on the need for transparency, accountability, and a proven model that accounts for rapid innovation, rather than rushing to legislate based on current imagination.

Funding and Future Outlook

  • ๐Ÿค Fathom is philanthropically funded by independent wealth, with a focus on societal good and avoiding industry influence or corporate funding.
  • ๐Ÿ”ฎ The speakers anticipate that the need for independent verification will become clear, leading to public-private partnerships and a focus on carrots over strict mandates.
  • ๐Ÿ“ˆ The aim is to create a flexible, adaptive framework that encourages innovation while ensuring AI aligns with public good and safety, preventing a scenario where it's too late to influence development.
Knowledge graph40 entities ยท 25 connections

How they connect

An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.

Hover ยท drag to explore
40 entities
Chapters18 moments

Key Moments

Transcript199 segments

Full Transcript

Topics14 themes

Whatโ€™s Discussed

AI RegulationSB 813Independent VerificationFathomAbundance InstituteAI GovernanceRegulatory LandscapeTort LawBenchmarkingAI SafetyPublic-Private PartnershipCalifornia SenatePhilanthropySocietal Good
Smart Objects40 ยท 25 links
Peopleยท 7
Companiesยท 9
Conceptsยท 17
Locationsยท 3
Mediasยท 3
Productยท 1