Joe Lonsdale, a small investor in Anthropic and cofounder of Palantir and 8VC, has accused leading AI companies of using warnings about existential risk to influence government policy and protect their market positions.
Speaking at Reuters Momentum AI in Austin on September 25, Lonsdale argued that people close to Anthropic and OpenAI are funding outside groups that favor stronger regulation. Reuters reported that he believes such advocacy could let a small number of powerful companies help write the rules and entrench an oligopoly.
The criticism came with important qualifications
Lonsdale did not dismiss safety as irrelevant. He said it is a real issue but argued that fear and regulation require a careful response. He also praised Anthropic’s technical team and performance, despite criticizing the policy direction associated with the company and some of its allies.
That dual position matters because Lonsdale has both financial and political interests in the debate. His venture firm invests in defense and technology companies, and he is a public advocate for a more permissive approach to innovation. His claim that safety advocacy is a market strategy is an interpretation, not proof of the motives behind every warning.
Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman and other technology leaders have recently called for slowing the development of the most capable systems if safeguards cannot keep pace. Current and former researchers have also raised concerns about loss of control. Those warnings can have competitive consequences without being fabricated for that purpose.
Regulation can favor incumbents and still address real risks
Complex licensing and testing rules often cost more for smaller companies to satisfy, which can strengthen established firms. The same rules may also reduce genuine hazards. Both effects can exist at once.
The policy question is therefore more specific than whether existential risk is real or invented. Lawmakers must decide which technical thresholds need oversight, whether rules apply evenly and who supplies the evidence used to set them.
Lonsdale’s comments sharpen the political conflict around AI safety: arguments about catastrophic risk are also arguments about market structure, government power and which companies get to define acceptable development.
