Does AI Need an Antitrust Exemption So It Doesn't Kill Everyone?

Does AI need an antitrust exemption so it doesn't kill everyone? That's the question hanging in the air after the CEOs of major AI companies asked to coordinate on safety without running afoul of competition law. The idea sounds reasonable until you remember those same companies face accusations of seeking regulatory capture and forming a cartel.

Context: From Arms Race to Calls for Rules

The AI industry is in a paradox: researchers at Anthropic and Google DeepMind have publicly resigned warning that models pose real threats, while companies ask for antitrust exemptions to coordinate on safety. All this happens as the Trump administration adopts a hands-off posture, leaving a regulatory vacuum that some compare to having cars but no lane markings, traffic lights, or speed limits.

Details: Kanter, Antitrust, and the AI Era

Jonathan Kanter, former head of the DOJ Antitrust Division, brings perspective. During his tenure, he led cases against Google, Apple, and Ticketmaster, and along with Lina Khan drove a reboot of American antitrust policy. Now a professor, he analyzes how AI regulation might interact with competition law. Companies argue they need to coordinate to avoid existential risks, but critics note they may be seeking regulatory cover ahead of IPOs and trying to fend off cheaper open-weight Chinese models.

Analysis: Coordination or Cartel?

The request for an antitrust exemption is suspicious. If companies truly fear AI could end humanity, they should welcome external oversight, not self-regulation. The risk is they use safety as an excuse to collude and block rivals. Kanter warns we need "basic rules of the road" for both companies and government. But the line between legitimate collaboration and cartel behavior is thin, and history shows industries rarely self-regulate well.

Implications: The Future of AI Competition

If the exemption is granted, it would set a dangerous precedent: tech giants could define the rules in their favor. If not, safety coordination might fail, leaving a fragmented landscape. Meanwhile, competition with China and open models adds pressure. The decision will shape not just safety but the structure of the AI market for decades.

The question isn't whether AI needs rules, but who writes them: the engineers building the models or the regulators tasked with protecting the public?

Source: The Verge