80,000 Hours Podcast

The 80,000 Hours team
undefined
35 snips
Nov 19, 2026 • 1h 47min

Julian Gewirtz on the AI risks that actually scare Beijing

Julian Gewirtz, a historian, China specialist, poet, and former senior US National Security Council official, examines how Beijing views AI risk. He discusses regime security, Anthropic’s influence on Chinese thinking, hidden AI incidents, diverging safety cultures, Taiwan and semiconductors, and the challenges of defining “safety” in US–China talks. He also explores whether incremental cooperation could overcome mutual distrust.
undefined
67 snips
Nov 16, 2026 • 4h 34min

39 specific projects to improve our chances with AGI | Holden Karnofsky, Anthropic

Holden Karnofsky, GiveWell and Open Philanthropy co-founder, now works on advanced AI risk at Anthropic. He explores 39 practical projects for improving humanity’s chances with AGI, from detecting deception and preventing model backdoors to security, model welfare, biosecurity, policy, and responsible scaling. He also examines rogue AI scenarios, frontier-lab incentives, and why AI safety work has become so tangible.
undefined
68 snips
Nov 13, 2026 • 19min

How the New Congress Can Save Us From AI Disaster | Rob Wiblin

With AI racing toward potentially transformative capabilities, Congress faces high hurdles to passing tough new rules. The discussion explores bipartisan steps on chip controls, model security, whistleblowers, funding, and oversight, then turns to bolder ideas: slowing recursive self-improvement, embedding government auditors, regulating compute, preserving state authority, and preparing legislation before the next AI crisis.
undefined
74 snips
Nov 12, 2026 • 15min

Is Anthropic on the cusp of RSI? Inside the redacted METR report | Luisa Rodriguez

An independent investigation inside Anthropic finds AI-assisted research accelerating dramatically, potentially bringing recursive self-improvement within reach. The discussion explores Anthropic’s own safety threshold, evidence that AI drives much of the speedup, and why computing, data, and algorithmic bottlenecks may not slow it down. It also examines report redactions, uncertainty in the findings, and calls for stronger safeguards and government limits.
undefined
78 snips
Nov 10, 2026 • 26min

The US is building the smartest AI. China is running a different race | Luisa Rodriguez

China’s top AI models trail America’s by only months despite having a fraction of the computing power. Explore how overseas cloud rentals, smuggling, gaming cards, and industrial-scale model copying help close the gap. The discussion also examines weak security, China’s broader deployment race, and whether tighter US policy could create crucial breathing room for AI safety.
undefined
72 snips
Nov 6, 2026 • 2min

Your AI crash course starts here: Our best episodes to get up to speed on transformative AI

A curated crash course on transformative AI, AGI, superintelligence, and recursive self-improvement. Topics span accelerated scientific progress, deceptive model training, economic and geopolitical upheaval, democracy, the US-China AI race, pandemics, AI consciousness, and what models may be thinking. The series also explores concrete projects and career paths for people eager to work in AI.
undefined
71 snips
Nov 5, 2026 • 2h 22min

An AI disaster hits. The president calls Xi. Then what? | Jeremie Harris (crosspost from NonZero)

Robert Wright, journalist and founder of NonZero, speaks with Jeremie Harris, AI entrepreneur and safety researcher. They explore why AI progress keeps accelerating while alignment lags, whether only an “AI Chernobyl” could change Washington’s course, and how OpenAI’s safety culture has evolved. They also examine US-China mistrust, crisis-driven cooperation, verification, and the alliances needed to prevent catastrophe.
undefined
44 snips
Nov 3, 2026 • 29min

3 new scandals show being a pain in the ass for AI labs is a legit job now | Rob Wiblin

Major AI companies face scrutiny over lobbying, researcher firings, and contracts restricting public criticism. The discussion explores how employees, journalists, campaigners, policy experts, and communicators can pressure these firms to improve. It also examines when insiders should speak up, resign, or organize external campaigns as the stakes and pace of AI development rise.
undefined
205 snips
Oct 29, 2026 • 2h 36min

#258 – Luke Muehlhauser on the most important AI safety work nobody is doing

Luke Muehlhauser, an AI governance funder and former Machine Intelligence Research Institute leader, explains why advanced AI is racing ahead of its safety infrastructure. They explore missing evaluators, incident investigators, auditors, and international verification bodies; warning shots reshaping policy; AI takeover, authoritarian lock-in, and biosecurity risks; and the urgent need for talented founders to build what’s missing.
undefined
85 snips
Oct 27, 2026 • 21min

Why Silicon Valley can’t slow down AI on its own | Luisa Rodriguez

AI developers worry the technology is advancing too fast, yet fierce competition keeps pushing the race forward. This discussion explores safety checkpoints, compute limits, restrictions on AI designing AI, and the possibility of a coordinated pause. It also examines how extra time could strengthen safety research, protect critical infrastructure, and let society shape the rules.

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app