Ep 828: Anthropic Responds: Why Claude’s CEO didn't sign the open model pact and the real reasons why
Everyday AI Podcast – An AI and ChatGPT Podcast
22 HOURS AGO
Ep 828: Anthropic Responds: Why Claude’s CEO didn't sign the open model pact and the real reasons why
Ep 828: Anthropic Responds: Why Claude’s CEO didn't sign the open model pact and the real reasons why

Everyday AI Podcast – An AI and ChatGPT Podcast
22 HOURS AGO
This podcast analyzes Anthropic CEO Dario Amodei's recent letter explaining why the company did not sign the Open Weights coalition supporting open-source AI models. The host argues that while safety concerns are cited, the primary motivation is financial, tied to Anthropic's upcoming IPO and its business model that relies heavily on per-token revenue, which is threatened by free, open-source alternatives.
The host contends that Amodei's letter is strategically aimed at Washington policymakers, not the public, to influence upcoming regulations. Anthropic's business is uniquely vulnerable because 80% of its revenue comes from token sales, a commodity being undercut by cheaper open-source models, especially from China. Unlike competitors like Google, Microsoft, and OpenAI, which have ecosystem lock-ins, Anthropic lacks a moat. The host points out that Anthropic's pre-IPO stock has dropped as the market shifts from token maximization to cost-per-task efficiency. Amodei's proposed solutions—chip bans, distillation crackdowns, and mandatory safety testing—are seen as a form of regulatory capture that would effectively ban open-source models, protecting Anthropic's proprietary business. The host notes contradictions in Amodei's stance, such as his own model escaping its sandbox, and argues that the letter mixes sincere safety concerns with commercial self-interest to preserve pricing power ahead of a public offering.
00:00
00:00
Anthropic's refusal is driven by IPO plans and revenue threats, not just safety.
04:19
04:19
Denies ever advocating banning open-weight models
06:39
06:39
Authoritarian governments could build more powerful AI than the US
12:14
12:14
All capable models should undergo mandatory safety testing
18:12
18:12
80% of Anthropic's revenue comes from token sales
20:00
20:00
Users cut costs by 95% through local setups
24:02
24:02
Anthropic's letter opposing open models is fear-based regulatory capture.
26:20
26:20
Safety concerns are both sincere and commercially self-serving.
27:13
27:13
Market shift from token maximization to cost-per-task efficiency
30:18
30:18
Anthropic bans open-source without saying the word.
36:32
36:32
Safety concerns serve IPO interests
37:37
37:37
Anthropic seeks regulatory help to protect its revenue model
