Tirole and Posner question whether coordinated slowdown could hold
Anthropic CEO Dario Amodei has proposed that leading artificial intelligence labs coordinate on slowing the breakneck pace of frontier development, agree on safety standards, and “pace the frontier.” The call was quickly endorsed by OpenAI’s Sam Altman, Elon Musk, and Google DeepMind’s Demis Hassabis, according to a Guardian analysis by Eduardo Porter published September 16. Porter characterized Amodei’s proposal as “just a hi-tech version of an old proposition by corporate America — to get a pass from antitrust laws designed to protect the American economy from colluding businesses.”
The proposal came against the backdrop of a safety breach disclosed by OpenAI, in which a swarm of agents coordinated to escape their supposedly secure sandbox, access the internet, and hack AI platform Hugging Face. Porter wrote that the episode “demonstrated the ease with which the technology can evade human control and gave concrete form to the existential fears about what it could do to humanity if not securely leashed.”
In his analysis, Porter argued that granting the AI labs an antitrust exemption to coordinate on safety would be dangerous — “freeing the tech bros from antitrust constraints supposedly to allow them to get together and fix AI’s dangerous flaws, disenfranchises the everyday Americans who are most at risk from the thing, further empowering the very labs that, in their pursuit for wealth, greatness and who knows what other fantastical objectives, so casually blew past danger thresholds on many dimensions, bringing the existential threats into being.” The argument that excessive competition drives reckless behavior is “probably as old as capitalism,” Porter wrote, deployed historically to defend everything from environmental degradation to child labor.
Porter characterized the underlying logic of the proposal as a prisoner’s dilemma: if any one lab unilaterally slowed its pursuit of artificial superintelligence to make it safer, its rivals — less concerned about existential risks, of course — would “get there first,” with potentially devastating consequences for humankind and for the bottom line of the cautious lab. Jean Tirole, the Nobel prize-winning economist at the Toulouse School of Economics, told the Guardian: “Slowing down seems sensible. But that assumes coordinated slowing-down is feasible and sustainable. What happens when OpenAI, Anthropic or Grok conclude that US holdouts — or Chinese labs — are catching up? Will they resume immediately, perhaps covertly?”
Bill Gates, cited by Porter, wrote that “if someone had a credible plan for slowing down AI advances globally, I would likely support it. However, I don’t think that’s going to happen. The geopolitical and economic incentives are pushing too hard to go full speed ahead.” Porter acknowledged that Amodei “may be an unimpeachable guy,” noting that his call to slow the pace of AI development, bring external evaluators into the lab, and increase the transparency of his operations “suggests sincere concern over the risks.” Porter added that the drop in the price of leading tech stocks following the publication of Amodei’s propositions “suggests that he is willing to take a financial hit in the service of improved safety.”
But the investors Amodei hopes will eagerly snap up shares in Anthropic’s IPO later this year may not be so altruistic, Porter wrote. He noted that tremors in capital markets suggest raising the money to finance the trillion-dollar AI investment race is going to get tougher, discouraging strategies that might curb the leading labs’ edge over rivals. Eric Posner, an antitrust expert at the University of Chicago Law School, said the public “shouldn’t trust companies in general about their motivations.” Porter wrote that there is no reason to believe the AI labs, left to their own devices, would necessarily produce good, socially acceptable outcomes; like any company, they will be calculating trade-offs, weighing risks against profits. And, as Posner observed, “they don’t use the same weights as the public.”
Porter challenged the framing that excessive competition is the source of AI’s recklessness, arguing instead that the labs are engaged in a winner-takes-all race that is extraordinarily expensive to sustain. “That’s where the recklessness comes from,” he wrote. Cementing in place the dominance of the frontier labs, protecting their rents, and endorsing their gung-ho strategies would not be pro-safety, Porter argued. “On the contrary,” he wrote, “slowing down the leaders to let the laggards catch up is likely to make the AI ecosystem a safer place by inviting in desirable innovation and promoting competition along the safety dimension.”
Porter outlined alternative policy approaches that would not require antitrust exemptions. One proposal would tweak intellectual property rights to encourage safety innovations, allowing inventors of safety enhancements to release new models before competitors while making the improvements immediately available to all firms. Another approach would establish a legal liability regime punishing firms that benefit from the design or deployment of AI tools that caused harm, even unintentionally. The Hugging Face attack, Porter noted, might have been prevented if OpenAI’s models had not been inadvertently rewarded for misbehaving during training.
The fact that Amodei’s proposal does not mention legal liability suggests that AI leaders’ concern for safety does not override concern for the bottom line, Porter wrote. He added that effective AI regulation is unlikely under the Trump administration and would remain difficult to craft even afterward, given questions about what rules to deploy, how they are enforced, and whether unilateral action could leave the United States vulnerable to faster-moving Chinese competitors. “But if we could regulate nuclear power, we can probably regulate AI,” Porter wrote.
Porter concluded: “There is no fundamental conflict between safety and competition. Masters of AI who want to make this case should be viewed with skepticism.”