Opinion: Anthropic Export Control Compliance Highlights Centralization Risks in AI, Decentralized AI Could Become Key Counterbalancing Force

BlockBeats News, June 14th, CoinFund founder Jake Brukhman stated that AI models naturally have centralizing attributes, making them a key target for government regulation and control. Anthropic's latest export control compliance action further confirms this trend.

He pointed out that decentralized networks can be a crucial counterbalance to the current situation, building sovereign, open, and public decentralized AI. The core challenge lies in the issue of computational power. Although it is widely believed that only trillion-dollar tech companies can afford cutting-edge model training, there is actually abundant global GPU computational power available. The key is to develop new distributed training algorithms.

Brukhman mentioned that teams such as Gensyn, Prime Intellect, Bagel, Pluralis, Nous Research, Macrocosmos AI, and Covenant AI have been exploring this direction. Although early on it was commonly believed to be infeasible, the reality has shown that distributed training can not only be achieved but also at a lower cost and with efficiency close to traditional solutions.

Furthermore, he believes that another major challenge facing decentralized AI is economic sustainability. While open-source models are important, they lack mature business models. Pluralis, by distributing model weights to participants, is exploring the commercial path of tokenizing AI models.

Brukhman stated that we are currently at a critical moment where AI will either move towards complete centralization, subject to scrutiny and unilateral government control, or establish a public AI on an open decentralized network, which will determine the industry's future direction.