Is the AI safety debate about safety or control?

Discussions have previously been held about creating a similar, federally run organization. However, the Trump White House — which includes vocal AI proponents and has largely pushed for an unregulated tech business (the administration has even sought to forcefully stop state governments from introducing their own AI laws) — hasn’t pursued anything. The administration’s “AI czar,” tech veteran David Sacks, has said that AI regulation should be left to the companies developing it.

Like the White House, prominent members of Congress seem equally disinterested in taking Amodei up on the offer to regulate his company. Not long after the Anthropic CEO’s essay was published, Speaker of the House Mike Johnson addressed concerns over catastrophic AI advancements thusly: “You’re not all going to be dead in 10 years,” he said.

A private organization would keep the industry unregulated except by voluntary commitments and would naturally bestow a certain amount of authority and influence to the companies responsible for creating it. This leads to another ongoing claim, which is that the industry’s calls for regulation are actually a strategy of “regulatory capture.” According to this argument, powerful companies that already enjoy a position of prominence in the industry may use regulatory stratagems to ice out or disadvantage smaller, less-resourced companies — thereby stifling their competition. This need not involve actual regulations but could involve voluntary industry standards that nevertheless put pressure on less-powerful firms.

Many prominent voices in the Chinese government have echoed that position. After Amodei called for “pacing the frontier” this week, China’s Ministry of Foreign Affairs spokesperson Guo Jiakun accused Americans of using “fearmongering” to disrupt the development of global AI governance, while an op-ed in China’s state-run newspaper called Amodei’s rhetoric straight out of a “Cold War playbook.”

Restructuring the way companies do business with China — one of the things Amodei calls for in his essay — could naturally be an opportunity for American companies to sway things in their own favor. Amodei even admits this is the goal. “If we execute these measures well, I believe they would slow China’s progress enough to widen America’s lead significantly over the next 3–5 years — the window when AI becomes geopolitically most important,” he wrote in his essay.

It’s not just China, however, that sees big American AI labs as seeking to throw the match in their own favor. Aidan Gomez, the CEO of Canadian AI firm Cohere, recently accused those companies of forming a “cartel,” writing, in a blog post: “AI needs guardrails. That is not the dispute and never has been. The dispute is over who writes them, who gets to participate and whose interests the rules are protecting.”

As Gomez notes, the field of AI safety is a deeply political one, wherein the terms by which safety are defined — and by whom — will ultimately help determine who has a head start and who gets left behind in the next stage of AI competition.