CHAPTERS
- 0:00 – 3:11
Trump dismisses AI risks as a “conspiracy” as industry calls for a slowdown
Kara and Scott react to Trump downplaying AI safety concerns and framing regulation as anti-innovation. They contrast that stance with Anthropic CEO Dario Amodei’s public push for independent safety evaluation and global standards, noting unusual support from Altman and Musk.
- •Trump argues AI needs only a “high IQ president,” not guardrails
- •Amodei calls for slowing development and adding independent safety evaluators
- •Push for global standards that include China
- •Speaker Mike Johnson signals interest but offers no concrete plan
- •Hosts question whether the ‘slowdown’ narrative is about safety or market power
- 3:11 – 6:15
The case for AI regulation: incentives, externalities, and why the U.S. is an outlier
Scott argues governments exist to balance innovation with the public costs that companies externalize. They compare the U.S. “unregulated trillion-dollar industry” experiment to China’s more restrictive AI rules and its higher public optimism.
- •Tension between economic progress and public harm is a centuries-old policy problem
- •U.S. frontier AI is effectively unregulated despite massive scale
- •China’s rules (bot disclosure limits, youth usage limits) as a contrasting model
- •Public sentiment: China optimistic vs. U.S. anxious without guardrails
- •Risk that U.S. delay becomes a major political issue
- 6:15 – 8:01
Liability and accountability: “AI won’t kill people—companies will”
The hosts argue that AI firms should be legally responsible for harms caused by their models and agents, including criminal liability in extreme cases. Scott uses analogies (dangerous dogs, unsecured firearms) to emphasize how liability changes incentives and behavior.
- •Analogy: owners prosecuted when dangerous dogs kill (Diane Whipple case)
- •Analogy: parents increasingly held responsible for unsecured guns
- •Executives framing AI as ‘out of control’ can be incentive-avoidance
- •Kara: stop anthropomorphizing AI; humans choose to deploy harmful tools
- •Call for liability up to and including jail for catastrophic outcomes
- 8:01 – 16:54
What a real AI oversight regime could look like (FAA/SEC/NTSB model)
Scott outlines a blueprint for oversight: pre-market safety review for sufficiently capable models, mandatory incident reporting, and independent evaluators with subpoena power. He argues governance must be funded by industry compute fees so oversight isn’t subsidized by taxpayers.
- •Need for a multilateral “IAEA-like” approach for frontier AI risks
- •Pre-market safety review for models above a capability threshold
- •Mandatory incident reporting modeled on the NTSB
- •Independent evaluator with subpoena power (not embedded/no-teeth reviewers)
- •Funding mechanism: transaction fee on frontier compute (FINRA-like)
- 16:54 – 17:50
Is the ‘slow down’ push strategic? Open source, consolidation, and narrative control
Kara and Scott argue that some leaders may be amplifying catastrophe narratives to justify regulations that entrench incumbents and constrain open source or smaller competitors. They warn that public backlash is building and the industry is unusually vulnerable to a high-profile failure.
- •David Sacks’ critique: slowdown advocacy may be self-interested, not altruistic
- •Fear narratives can create demand and reinforce dominance (Huang quote)
- •Industry vulnerability: one serious incident could force shutdowns
- •Media and public ‘waiting’ for a rogue-bot headline
- •Populist anger shifting from data centers to model safety and corporate power
- 17:50 – 22:46
OpenAI delays IPO talk while Anthropic reportedly lines up Nasdaq
The conversation turns to money and timing: Altman says it’s a bad moment to go public due to safety concerns, while Anthropic is reported to be preparing for an IPO. Scott frames both CEOs as squeezed between societal risk, investor pressure, and the absence of clear regulatory requirements.
- •Altman confirms OpenAI won’t go public this year; Kara calls it an excuse
- •Scott revisits his earlier prediction that an OpenAI IPO wouldn’t happen soon
- •Anthropic reportedly chooses Nasdaq and could IPO as soon as October
- •Investor/employee liquidity pressure vs. reputational and systemic risk
- •Markets react sharply to extinction-risk commentary; governance vacuum amplifies fragility
- 22:46 – 26:39
California’s new child online safety and AI chatbot rules
Kara highlights Newsom’s package of bills targeting addictive social media features for minors and stronger safeguards around AI companion chatbots. They debate the downsides of state-by-state patchwork but argue Congress has failed to act and public demand for protection is rising.
- •Ban on addictive platform features for users under 16
- •Stronger safeguards and penalties tied to AI companion chatbots
- •Fines up to $1M per child if negligence is found
- •Patchwork regulation vs. need for federal standards
- •Prediction that political pressure will intensify as harms become clearer
- 26:39 – 30:03
Mark Carney courts Europe as Trump antagonizes Canada
After the break, they discuss Canada’s Mark Carney seeking closer ties with the EU—potentially even associate membership—as a strategic response to U.S. trade hostility. Scott argues this is a major opening for democratic alliances and a long-term loss for the U.S. if it continues to bully allies.
- •Carney explores EU alignment on energy, AI, minerals, defense, and visas
- •Scott pitches a broader alliance of democracies (trade + diplomacy)
- •EU governance challenge: 27 vetoes makes unified policy difficult
- •Canada has more in common with EU than U.S. under current posture
- •Trade war creates opportunity for non-U.S. suppliers to capture business
- 30:03 – 35:02
Why business-style domination fails in geopolitics: alliances over M&A thinking
They argue U.S. leaders are treating diplomacy like a leverage game, undervaluing bureaucrats and diplomats while idolizing business tactics. Scott and Kara emphasize cooperation as the key advantage of democracies and warn that alienating allies amounts to “superpower suicide.”
- •Critique of “we don’t need anyone” mentality despite interdependence
- •Government requires coalition-building, compromise, and long-term strategy
- •Historical examples: rebuilding former enemies into allies
- •Warning that tariffs and bullying are economically self-destructive
- •Kara underscores cowardice and complicity among those surrounding Trump
- 35:02 – 43:39
Mitch McConnell returns frail—renewed push for term limits and age gating
Kara and Scott discuss McConnell’s return to the Senate after hospitalization and rehab, arguing that extreme age and cognitive decline undermine governance. They advocate for term limits, mandatory physicals/cognitive testing, and structural rules that reduce the burden of stepping aside.
- •McConnell returns in a wheelchair, aims to finish the farm bill and back Ukraine
- •Scott: cognitive decline includes inability to recognize decline
- •Personal stories about taking away an elderly parent’s driver’s license
- •Comparisons to Feinstein and broader leadership gerontocracy
- •Policy ideas: age gating, mandatory annual physicals, clearer retirement norms
- 43:39 – 53:37
Wins & Fails: corruption report, ‘Practical Magic 2,’ and Newsom’s child-safety package
In closing, Kara flags a ProPublica report about a Russian oligarch allegedly helping pay for Donald Trump Jr.’s wedding and celebrates the box office success of ‘Practical Magic 2.’ Scott names Newsom’s child safety laws as his win and points to the broader failure of AI governance—an industry ‘grading its own homework’—as his fail.
- •Kara fail: alleged oligarch-funded Trump Jr. wedding as emblem of corruption
- •Kara win: ‘Practical Magic 2’ box office success and cult-classic renaissance
- •Scott win: California online child safety bills and AI companion safeguards
- •Scott fail: absence of an AI regulator akin to FDA/FAA; self-regulation is insufficient
- •Both: governance vacuum increases systemic risk and public backlash potential
