CoinFund CEO flags government intervention in Anthropic model access

Anthropic disabled the specific models to comply with a government order restricting access for foreign nationals, whether they are based inside or outside the United States. While the company maintains that other Claude models remain available, the move has ignited fresh debate regarding the concentration of power in frontier AI. Brukhman contends that AI models currently function as a centralizing force, making them prime targets for unilateral state control and potential censorship.

To mitigate this, industry players are shifting focus toward distributed GPU training and tokenized model ownership. Brukhman highlighted companies like Gensyn, Prime Intellect, and Pluralis as key entities attempting to prove that decentralized training is both cost-effective and efficient enough to challenge centralized systems. Pluralis, in particular, is experimenting with splitting model weights among network participants, a structure designed to bypass the need for a single, vulnerable central authority. As massive capital continues to flow into centralized infrastructure—such as the $36 billion in debt financing lined up by Blackstone and Apollo for Google TPU expansion—the divide between private, restricted models and open, public networks is widening.