The AI ‘Slowdown’ Is an Antitrust Mess

AI companies' public discussion of development slowdowns risks antitrust scrutiny and potential years-long investigations.

Leading AI companies, including OpenAI and Anthropic, have publicly discussed a potential "slowdown" in AI development. However, antitrust experts warn that such collective action, especially if framed as an agreement to reduce output, could violate US antitrust laws like the Sherman Act. The phrasing used by company executives is crucial, as regulators scrutinize language that might imply anticompetitive collusion. While the companies' stated goal is to ensure advanced AI models do not become dangerous or "rogue," an explicit agreement to decelerate could be interpreted as an illegal pact to limit trade. Experts suggest framing such efforts around safety protocols and risk mitigation, with a slowdown in releases being a natural consequence, rather than an agreed-upon objective. Meta CEO Mark Zuckerberg, for instance, emphasized the natural market incentive for AI agents to behave well to avoid alienating consumers.

The rhetoric surrounding an AI "slowdown" presents a significant antitrust challenge, as a coordinated pause in development could be misconstrued as an anticompetitive agreement to reduce output under laws like the Sherman Act. Antitrust experts note that the language used by company leaders is critical, with a collective agreement to "take it easy" potentially triggering investigations. Companies might face allegations of "quality fixing" if they collectively agree not to improve products beyond necessary safety measures, drawing parallels to past antitrust cases involving industries like automotive emissions technology. While self-regulation and industry standards are possible, regulators often view such exemptions with skepticism, fearing they could entrench dominant players.

The market and regulatory landscape for AI development is becoming increasingly complex, with companies facing pressure from both political and competitive forces. While some AI firms express genuine concerns about safety and the pace of innovation, there's suspicion that discussions of a slowdown could be a strategic move to consolidate market position, particularly as major players like OpenAI and Anthropic approach IPOs. The potential for lengthy "conduct investigations" by antitrust authorities means that even a perceived anticompetitive agreement could lead to substantial legal and operational burdens, forcing companies to navigate these challenges carefully.