scripod.com

#221: Anthropic vs. the White House, Microsoft CEO on the Future of Firms & AI’s Token Crisis

Shownote

Anthropic's most powerful models are still offline, and the U.S. government now wants a guarantee no lab can give. Paul Roetzer and Mike Kaput unpack the ongoing export-control standoff, the Lutnick letter, and what it means for the models expected this we...

Highlights

This episode delves into the escalating standoff between Anthropic and the U.S. government over export controls, which has left powerful AI models offline. The hosts also explore the implications of Microsoft's CEO vision for the future of work, the chaotic state of AI pricing, and a wave of talent moves between major AI labs. Additional topics include the G7's push for AI governance, Midjourney's expansion into medical imaging, and new research demonstrating AI's superhuman persuasive abilities.
02:58
42% feel troubled by Anthropic's model pull.
10:40
Claude preview significantly improved exploit development
25:57
AI serving people, not replacing them.
37:00
Token costs explode with agentic AI
1:01:52
Demis is mission-driven and may leave if productization wins.
1:04:57
AI leaders called for a U.S.-led coalition to set AI rules.
1:09:34
AI-driven medical advancements could reduce deaths and healthcare costs.
1:13:00
AI can out-persuade expert humans in conversation.
1:17:01
AI can critique your speaking style by analyzing transcription, audio, and video.
1:22:17
SpaceX acquired AI startup Cursor for $60B

Chapters

Intro
00:00
Anthropic vs. the White House
04:44
Microsoft CEO on the Future of the Firm
22:37
AI Pricing and Usage Strategy
34:37
Noam Shazeer Joins OpenAI (and Other Major AI Hiring Updates)
56:04
Trump's G7 AI Push
1:04:57
Midjourney Launches a Medical Division
1:09:33
AI Can Now Out-Persuade Expert Humans
1:12:59
AI Use Case Spotlight
1:17:00
AI Product and Funding Updates
1:22:12

Transcript

Paul Roetzer: I could actually see this administration banning the use of Chinese open source models by US companies within the next 30 days. Because what's happening now is to save on cost, companies are just going to start using DeepSeek models and thing...