White House warns of 'regulatory capture,' 'cartel'

Startups slam moves as barriers to entry for latecomers

OpenAI, Anthropic, Google already forming coalition

Elon Musk of SpaceX AI, Dario Amodei of Anthropic and Sam Altman of OpenAI have all called for slowing AI development and putting safety guardrails in place, while actively shaping the regulatory framework. Critics say the push is designed to erect barriers that disadvantage latecomers. [AFP]
Elon Musk of SpaceX AI, Dario Amodei of Anthropic and Sam Altman of OpenAI have all called for slowing AI development and putting safety guardrails in place, while actively shaping the regulatory framework. Critics say the push is designed to erect barriers that disadvantage latecomers. [AFP]

AI startups are pushing back hard against calls by industry leaders Anthropic and OpenAI to slow the pace of AI development, accusing the dominant players of trying to craft regulations that suit their own interests while blocking smaller rivals from catching up.

Bloomberg reported Tuesday (local time) that executives at AI-related startups fear the "AI slowdown" proposals and "safety standards" championed by the top companies will function as regulatory barriers that shut out smaller, later-stage competitors.

Faraj Alaei, CEO of Cogni Chip, which is developing models to improve semiconductor design efficiency, said regulatory barriers "always favor incumbents" and amount to "building a wall after you've already passed through." Naveen Rao, CEO of AI computing startup Unconventional AI, warned that the greatest risk is limiting the diversity of effort within the industry, saying he does not want "a monoculture where Anthropic in effect runs every application." Aidan Gomez, CEO of Canadian AI startup Cohere, wrote in a blog post that leading AI companies risk forming a kind of "cartel," arguing that "these oligopolists are now asking to be allowed to distort the rules of competition and set the terms for everyone else."

The criticism echoes remarks by David Sacks, the White House science and technology adviser, who said AI companies are "volunteering to be regulated" while seeking "their preferred regulatory framework in return" — a dynamic he described as "regulatory capture," in which the leading firms among those subject to oversight effectively choose the rules that govern them. Startups worry that if the Big Tech companies driving the debate also shape the regulatory response, the resulting rules will favor incumbents and disadvantage challengers.

Earlier, Anthropic CEO Dario Amodei on Saturday proposed that AI model developers "voluntarily cooperate to set standards" and slow the pace of technological advancement to reduce the risk of catastrophic harm — while also calling for regulations that would cover even companies that refuse to cooperate. Amodei proposed that the US government provide a "narrow waiver" allowing companies to coordinate on safety standards without antitrust scrutiny, a position that OpenAI and SpaceX AI immediately endorsed.

Federal Trade Commission Chair Andrew Ferguson pushed back at an event Tuesday, saying that when companies simultaneously demand sweeping regulations and antitrust exemptions, "alarm bells go off," and accusing AI companies of seeking "entry barriers" to block competitors. Sacks also said the coordinated push by leading AI firms to shape regulatory design "looks like a cartel."

Gil Luria, head of technology research at D.A. Davidson, said Silicon Valley has run this play before. "They use regulatory ambiguity to grow their businesses, and once they're big enough, they kick away the ladder and embrace regulations large enough only they can handle and control," he said. The concern is that rules formed after the industry has already matured will fall disproportionately on startups — the very companies that could one day challenge the incumbents.

Even as the criticism mounts, coordination among the leading AI companies is becoming more organized.

Chris Lehane, OpenAI's chief global affairs officer, said Tuesday that an antitrust exemption is unnecessary, adding that OpenAI has already been working with Anthropic and Google on AI safety cooperation. Anthropic, OpenAI and Google are separately reported to be in discussions about establishing a joint body to set common AI safety standards.

SV Angel, a venture firm that has backed both Anthropic and OpenAI, has hired Jay Carney — who served as White House spokesperson under the Barack Obama administration — to launch "Project Blueprint," an initiative to jointly design AI policy including governance frameworks for frontier models. Apparently aware of the concerns, Carney said he is "not working for the agenda of any particular company" and that "every part of the AI world should have a voice." Both Anthropic and OpenAI have expressed support for Project Blueprint.


kate01@heraldcorp.com