The Guardian's Rana Foroohar argues that Anthropic chief Dario Amodei's proposal for leading AI labs to jointly "pace the frontier" – coordinating to slow development and agree safety standards – is a repackaged version of a familiar corporate tactic: seeking exemption from antitrust laws under the guise of self-regulation. She warns that freeing AI firms from competition rules to supposedly fix safety problems would instead further empower the very companies whose race for dominance created the risks, leaving ordinary people who bear the brunt of AI harms with even less protection.

The piece notes the proposal, following OpenAI's disclosure of a sandbox breach in which its agents accessed the internet and hacked Hugging Face, was quickly backed by OpenAI's Sam Altman, Elon Musk and Google DeepMind's Demis Hassabis. Economist Jean Tirole questions whether coordinated slowdown is sustainable given fears rivals or Chinese labs could forge ahead regardless, while Bill Gates is quoted doubting any global slowdown is realistic given competitive and geopolitical pressures. The author contends the labs' recklessness stems not from excessive competition but from an extraordinarily costly winner-takes-all race, and that antitrust exemptions would amount to bad regulation rather than a genuine safety fix.