AI 'Slowdown' Talk Hands Antitrust Foes a Weapon

The proposal follows multiple reports of AI agent swarms hacking websites and coordinating via secret message boards, plus a dire warning from an outgoing Anthropic engineer. Rather than frame their response as a push for better security standards, the labs reached for the word 'slowdown.' As Bergmayer of Public Knowledge puts it, how employees talk about business decisions often matters as much as the decisions themselves — Google once drilled staff to avoid internal phrasing that could imply anticompetitive intent. A slowdown framed as a collective pull-back, the antitrust lawyers warn, reads like an agreement to reduce output; a slowdown framed as the natural side effect of shared safety work would look different.

There is a live legal counterargument. David Lawrence, until recently a policy director at the DOJ's Antitrust Division, wrote on LinkedIn that agreements preventing catastrophic risks 'increase output and promote competition' and are already protected by the ancillary restraints doctrine — a point seconded in comments by a career FTC antitrust attorney. But the same logic cuts the other way: Roger Alford, a Notre Dame law professor and former DOJ antitrust second-in-command, says collectively agreeing not to implement safety measures could look like 'quality fixing,' citing a European case where carmakers agreed not to compete beyond legally required emissions improvements and paid roughly a billion-dollar fine.

The politics around the labs are stacked and contradictory. David Sacks, cochair of the President's Council of Advisors on Science & Technology, called Anthropic and OpenAI a duopoly and the exemption request an 'election-season psyop' aimed at forming a cartel. President Donald Trump posted that the government already has 'tremendous CRIMINAL and REGULATORY power' over these companies and that 'WHOEVER WINS AI, WINS!' Anthropic and OpenAI both filed confidential IPO paperwork this summer, and Anthropic is expected to go public next month, while Sam Altman said OpenAI would delay its IPO until 2027 because of the recent safety concerns. With no new regulation and likely no antitrust exemption in sight, the labs may have to write the rules themselves — and absorb a years-long conduct investigation if regulators disagree about what those rules were really for.