Don't Worry About the Vase Podcast

A deep dive into Anthropic's Responsible Scaling Policy v3.0, focusing on how the new rules work and what practical levers remain. Discussions cover risk-report cadences, sabotage and insider-threat models, and capability thresholds for chemical and biological dangers. The conversation also examines removed safety checks, roadmap timelines, and who gets to decide what counts as a "strong argument."

Conversations range from leaks and naming controversies around a major AI model to how visual benchmarks can be gamed. They cover cybersecurity threats from supply-chain attacks and model-enabled offense. Topics include agents wrecking cold outreach, debates over job displacement and life after work, hardware shortages, and lively speculation about superintelligence secrecy.

Sam Bowman, an Anthropic researcher and commentator, gives a concise take on RSP v3 and why transparency and reporting matter. The conversation covers the policy shift from hard barriers to aspirational safeguards. Short, sharp scenes debate honesty, trust, broken promises, strategic racing, and whether clearer accountability really changed outcomes.

A spoiler-filled take on a documentary about advanced AI, its existential risks, and the filmmaker's emotional journey from panic to seeking optimism. Short, clear explainers cover instrumental convergence and why some experts worry. Discussion also dives into deepfakes, job disruption, tech race dynamics, leadership responses, and calls for public action and treaties.

Conversations jump from the OpenAI Foundation’s limited safety pledges to Congressional AI guardrail proposals and legal loopholes. They unpack US-China tussles, chip smuggling, and data center water scares. Debates swirl around pausing frontier models, verification for international deals, metagaming in model training, and whether systems show long-term goals. Public protests and political rhetoric round out the discussion.

A court’s preliminary ruling in Anthropic v. DoW takes center stage. Testimony and emails that undercut the DoW’s sudden designation get parsed. Technical clarifications about model deployment and update mechanics are explained. Strategic and legal debate surfaces over narrow remedies, foreign-worker claims, and immediate business harms. A judge’s pointed rebuke and its practical effects close out the discussion.

Quoted/Guest Contributor provides recurring sourced interjections and cited remarks. They riff on language model uses, evaluation quirks, agent risks and phone-call demos. Conversation touches on media generation, hiring shifts at AI labs, diffusion and economic effects, and Anthropic’s 80,000-interview survey about hopes and worries around AI.

Quoted contributors like Adam Wolf, Dean W. Ball and Gary Tan offer short takes on Anthropic updates. They unpack Claude Code Auto Mode, Claude Cowork controlling keyboard/mouse/screen, and risks of full computer access. Conversation covers agentic coding tradeoffs, code review automation, Dispatch workflows, and fast product iteration with practical tips and cautionary notes.

A probing critique of Socratic methods, from refutation tactics to whether inquiry needs new data. Debates about rubber ducking, coherent extrapolated volition, and when algorithms count as thinking. Tangents on politics, persuasion, equality, and whether Socratic love and virtue capture real human goods. Ends with reflections on preparing for death and the value of memorial practices.

A long, critical take on Callard’s claim that two-person Socratic dialogue is the royal road to wisdom. Debates whether constant inquiry should trump practical life and when not to philosophize. Critiques Socratic tactics, stories, and intellectualism. Surveys rival ethical theories and explores consequences for law, motivation, and meaning.