The Rhetoric Trap: How AI Labs Are Coding Their Own Antitrust Nightmare

In the high-stakes race for artificial intelligence dominance, the language used in the boardroom is often as consequential as the code written in the lab. As first reported by Wired, leading AI companies have recently called for a coordinated development “slowdown” following reports of AI agent swarms hacking websites and a dire warning from a departing Anthropic engineer. While the labs may view this as a necessary safety pause, they are inadvertently walking into an antitrust minefield.

**Opinion:** The strategic error here is not the desire for safety, but the framing. By utilizing the specific term “slowdown,” these companies have shifted the conversation from technical standards to output reduction—a primary red flag for antitrust regulators. In attempting to signal caution, they have provided a linguistic roadmap that allows the government to treat their coordination as a cartel.

According to John Bergmayer, legal counsel for the nonprofit Public Knowledge, the phrasing of these decisions is critical. Bergmayer notes that economists typically examine whether companies are reducing output or making a pact to “take it easy.” By framing the shift as a “slowdown” or a “pause” rather than a collaborative effort to establish security protocols, the industry has boxed itself into a corner where its actions could be interpreted as an anticompetitive agreement to reduce trade.

There is a safer path the industry could have taken. Bergmayer suggests that if the labs had emphasized the development of safety protocols to prevent catastrophic risks, the resulting decrease in model releases would have appeared as a natural side-effect rather than a collusive agreement. Meta CEO Mark Zuckerberg echoed this sentiment, avoiding the “slowdown” terminology entirely. Zuckerberg argued that companies have a “strong natural incentive” to solve “misalignment”—the industry term for models acting against human intent—because those who fail to do so will lose their competitive edge.

From a legal standpoint, the labs may have more protection than their own rhetoric suggests. David Lawrence, a former policy director for the Department of Justice’s Antitrust Division, noted on LinkedIn that agreements designed to prevent catastrophic risks actually promote competition and increase output, falling under the “ancillary restraints doctrine.” Roger Alford, a Notre Dame Law School professor and former second-in-command at the DOJ Antitrust Division, indicates that a different path—colluding to avoid safety measures—could trigger accusations of “quality fixing.” Alford cites a European case where car companies were fined roughly a billion dollars for agreeing not to compete on emissions-reducing technology beyond legal requirements.

Despite these legal defenses, the optics remain volatile. David Sacks, cochair of the President’s Council of Advisors on Science & Technology, has been explicitly critical, accusing OpenAI and Anthropic of operating as a duopoly. Sacks characterized the request for an antitrust exemption as an “election-season psyop” and a transparent attempt to “form a cartel.”

Ultimately, the industry is grappling with a tension between market pressure and existential risk. Bergmayer observes that while some companies use regulation to “pull up the ladder” behind them, employees are genuinely concerned about the pace of development. However, by failing to utilize existing frameworks—such as the National Cooperative Research and Production Act of 1993, which allows for standards-development organizations via FTC and DOJ notification—the AI labs have traded a structured regulatory path for a rhetorical disaster.