「Anthropic's Claim of Becoming the Sole Company」 Sparks Debate, Ray Dalio Responds: AI Inherently Tends Toward Centralization, Regulation is the Solution

According to Dynamic Beating's monitoring, investor Gavin Baker claimed on the All-In Podcast that several people he trusts told him that Dario Amodei once said Anthropic could one day become the "world's only private company." Anthropic researcher Sholto Douglas promptly denied this, stating that the claim was "completely false" and that Baker's sources were lying.

Baker acknowledged the clarification but continued to inquire. He noted that the rumor gained traction because it aligned with Dario's long-standing public stance: AI is very dangerous, power could be highly centralized, and therefore, regulation is necessary. Baker, on the contrary, believes that the more dangerous AI becomes, the more its capabilities should be decentralized, rather than entrusted to a few companies and governments.

Subsequently, Dario personally responded but did not address the "sole company" statement further. He argued that AI would inherently tend towards concentration among a few giants due to chips, computing power, and economies of scale. Open sourcing weights would only allow more people to get models but wouldn't solve who has the capacity to train the most powerful models. Therefore, his proposed solution is regulation, focusing on the strongest entities: rigorously scrutinizing cutting-edge models, with less regulation for small companies and followers in pursuit. While open sourcing weights could be maintained, once the capabilities approach the cutting edge, they should also be scrutinized. Baker is concerned that regulation would foster centralization, while Dario views regulation as a mechanism to counterbalance centralization.

He also addressed Baker's criticism of his constant emphasis on AI risk. Dario stated that the public's lack of trust in AI companies is not because risk is overemphasized but because the industry overpromises and underdelivers. Endless promises of "AI can cure cancer" are ineffective; actually curing cancer is what matters. Anthropic is intensifying its focus on biology and healthcare, hoping to see some early results in the coming months.