AI Panic: Dario’s Warning, Trump’s Dismissal, and OpenAI’s IPO Delay

The episode opens with Kara recounting a conversation with San Francisco Mayor Daniel Lurie after Scott made critical comments about the city's homelessness crisis. Lurie responded thoughtfully with data showing improvements in crime and homelessness metrics, exemplifying effective governance focused on operations rather than ideology.

The main focus shifts to AI regulation. Anthropic CEO Dario Amodei published an essay calling for the AI industry to slow development, implement independent safety evaluations, and establish global standards including China. This stance is backed by Sam Altman of OpenAI and Elon Musk. However, Trump dismissed these concerns as a "sick conspiracy" against AI and data centers, claiming only a "strong and smart, high IQ president" can provide necessary oversight.

Scott argues that current regulation is inadequate and proposes a framework modeled after the FAA and SEC: pre-market safety reviews for models above certain capability thresholds, mandatory incident reporting similar to the NTSB, funding through transaction fees on frontier compute, and independent evaluators with subpoena power. He criticizes David Sachs's observation that the slowdown motivation is partly about protecting market dominance through stifling open-source competition, though acknowledges Sachs has been irresponsible in blocking reasonable guardrails.

Regarding OpenAI's IPO delay, Sam Altman confirmed the company won't go public this year, citing safety concerns. Kara notes this creates tension: executives face pressure from investors expecting returns while also bearing responsibility for potential harms. Anthropic, by contrast, is preparing for an October IPO despite Amodei's warnings.

Kara emphasizes that the trillion-dollar AI industry remains entirely unregulated—unprecedented for an industry of this magnitude. Scott stresses that companies must be held liable for their AI systems' actions, comparing it to liability for dangerous animals or unsecured firearms. Both argue the Trump administration's incompetence and corruption make this worse, with RFK Jr. as HHS head showing no interest in investigating biosecurity risks from frontier models.

The conversation then shifts to Mark Carney, Canada's new prime minister, who is pursuing associate membership with the EU rather than full commitment to a US-dominated North America. Scott views this as strategic positioning given Trump's tariff threats and bullying approach to trade. Carney represents dignified leadership that recognizes interdependence rather than dominance.

Mitch McConnell's return to the Senate in a wheelchair prompts discussion about age-gating in politics. Scott argues that cognitive decline makes self-assessment impossible and advocates for mandatory physicals for elected officials, similar to airline pilot requirements. Neither should have served beyond reasonable age limits.

Finally, Governor Newsom's package of 13 online safety bills is praised as exemplary governance filling a federal void. The laws ban addictive features for users under 16, require AI chatbot risk assessments, and impose penalties up to $1 million per child for negligence. Scott notes Newsom is effectively acting as president on tech regulation while managing economic and political pressure from the industry.