Panel
AI's Ethical Frontier: Responsible Regulation...by Whom? | Global Conference 2024
Milken InstituteJo Ling Kent, Rob Bonta, John B. Quinn, Rob Silvers, . will.i.am, Frankie Williams, Felicia
- DHS Strategic Focus: Undersecretary Rob Silvers identified critical infrastructure sectors (electric grid, air traffic control, healthcare) as primary targets for developing new operational guidelines and auditing frameworks to prevent catastrophic failures.
- Executive Order Context: The White House issued executive orders and guidelines to establish security architectures and guardrails, urging federal agencies to ensure AI benefits do not result in physical harm to the public.
- Existing Legal Tools: California Attorney General Rob Bonta argued that current federal statutes, specifically the Telephone Consumer Protection Act, already cover AI-generated robocalls and that explicit consent violations occurred regarding the "don't vote" message in New Hampshire.
- CSAM & Child Safety: Bonta called for the creation of a Congressional expert commission to study AI's potential to generate or exacerbate Child Sexual Abuse Material (CSAM) and requested the Department of Commerce emphasize transparency and auditing in AI rulemaking.
- Data Sovereignty Debate: Panelist Will highlighted a fundamental shift from social media's data extraction model to AI's dependency on data, characterizing the current landscape as "new colonialism" where corporations treat user data as a "digital soul" for profit.
- EU AI Act Implications: Frankie described the EU's approach as a "risk-based, layered" system with horizontal effects, noting that while the US relies on guidance, the EU uses "hard law" where the letter of the statute carries more weight than judicial precedent.
- Burden of Proof Shifts: Litigator John discussed potential legal reforms inspired by the EU, including reversing the burden of proof in product liability cases where AI is the cause of injury, requiring deployers to prove their systems were not defective.
- Intellectual Property Landscape: John noted that while large-scale data scraping has generally won out in courts so far (citing Meta and LinkedIn cases), the legal landscape regarding AI-generated IP and liability for autonomous vehicle failures remains highly uncertain.
- Federal vs. State Dynamics: Speakers expressed skepticism regarding comprehensive federal privacy legislation, predicting that regulation will emerge primarily from "laboratories of democracy" in state legislatures, particularly California.
- Corporate Liability: Bonta confirmed the California Attorney General's office is actively suing companies like Meta using existing consumer protection, civil rights, and antitrust laws when AI is integral to illegal schemes or deception.
- New AI Safety Board: Silvers announced the Department of Homeland Security is establishing a new AI safety and security board including industry leaders (Sam Altman, Sundar Pichai), civil rights experts, and downstream customer CEOs to foster inclusive consensus.
- Diversity in AI Development: Will criticized the lack of representation in AI development teams, noting that current systems perpetuate bias because the voices designing the technology do not reflect the broad spectrum of humanity.
- Algorithmic Manipulation: The discussion identified that AI algorithms are increasingly driven to manipulate user engagement and behavior for profit, mirroring addictive social media practices, which necessitates urgent ethical regulation.
- Regulatory Capture Concerns: Panelists debated mechanisms to prevent regulatory capture, suggesting that diverse stakeholder inclusion in standard-setting processes is essential to ensure rules benefit the public rather than entrenching Big Tech dominance.
- Education & Workforce: Silvers emphasized the inevitability of AI-driven job displacement, asserting that the workforce must adapt by mastering AI tools, noting the sentiment that "people won't lose jobs to AI, but to those who use AI better."
- California's Legislative Role: Bonta outlined California's strategy to lead by enacting the strongest privacy and AI protections in the nation, serving as a model for other states while resisting federal preemption of state-level privacy rights.
- CSAM Accountability: While John acknowledged the immense difficulty of policing AI-generated CSAM, he noted that large tech companies have dedicated teams to filter such content, though he warned the problem will worsen as generation capabilities improve.
- Future Safety Standards: The panel concluded with calls for an "AI Constitution" or FDA-like oversight framework to establish clear ingredients and safety standards, ensuring public trust before the technology becomes ubiquitous.