AI Regulation Controversy: Industry Conflict Comes to the Surface

The controversy over AI regulation has emerged into the public arena on social media. Three figures—investor Gavin Baker, former White House advisor David Sacks, and Meta researcher Yann LeCun—publicly criticized Anthropic CEO Dario Amodei by name. Their shared claim: 'He is stoking fear about regulation to create a regulatory environment favorable to his own company.'

In response to this criticism, Amodei presented a counterargument. He expressed the position that regulation is not merely about limiting corporate power but can serve as a means to restrain the concentration of power itself in AI development. Moreover, he pointed out that simply releasing open-source AI models—those anyone can freely use—does not change the fundamental nature of the problem. He presented the view that openness does not distribute power; rather, it shifts power to those controlling the most computational resources: corporations or governments that own high-performance chips and large-scale data centers.

Behind this controversy lies a longstanding structural debate in the AI industry: 'open versus closed.' The open-source camp has argued that publishing models democratizes technology and prevents power from concentrating in the hands of a few major companies. Meanwhile, those prioritizing safety worry about the risks of powerful AI models being widely released. This controversy can be seen as a manifestation of this conflict becoming entangled with business interests, surfacing as a struggle over the specific design of regulation.

The point Amodei raises about 'computational resource concentration' is technically significant and cannot be overlooked. Current AI development relies on training using vast quantities of GPUs (graphics processing units), and the cost of procurement and scale required are said to be achievable only by major corporations or government institutions. Even when open-source models are released, only those possessing computational resources can meaningfully leverage and improve them at scale—a structural problem Amodei is highlighting.

The points made by the three critics also carry weight. It is a well-known concern in technology policy discussions that regulation tends to restrict competition and create 'barriers to entry' that protect existing large corporations. Given that Anthropic has already conducted large-scale fundraising and established a certain position in the industry, the suspicion that 'those wanting regulation benefit' is not entirely without basis.

What this conflict reveals is the reality that the debate over AI regulation cannot be separated from a power struggle over who will lead the industry—it goes well beyond a simple 'safety versus freedom' dichotomy. In future regulatory discussions, how to position computational resource concentration, not just the scope of model disclosure, will likely be a key focal point. This is a debate where technology, policy, and business are intricately entangled, and it warrants continued attention.