Nothing gets a room’s attention like accusing the smartest AI labs on Earth of playing chess with the rulebook. That’s essentially what venture capitalist Joe Lonsdale did on stage in Austin, Texas, this week – and he did it while still holding a stake in one of the companies he criticized.
His message was blunt. Fear sells. And in Washington, fear can also buy influence.
Table of Contents
Who Is Joe Lonsdale?
Joe Lonsdale co-founded Palantir Technologies, the data analytics firm known for its government and defense contracts. He also runs 8VC, a venture capital firm with investments across tech and defense. Notably, Lonsdale is an investor in Anthropic, the AI company behind Claude.
That last detail matters. He isn’t an outside critic taking potshots. He’s an insider questioning the playbook of a company he backs financially.
What Happened at the Reuters Momentum AI Austin Event
Lonsdale made his comments at the Reuters Momentum AI Austin event on September 25, 2026. He took direct aim at people connected to Anthropic and OpenAI who fund outside advocacy groups pushing for tighter AI regulation. U.S. News & World Report
His concern wasn’t that these companies talk about AI risk. It’s who benefits when the rules finally get written.
“We don’t want an oligopoly that controls all of our policy here with the government,” Lonsdale said, describing the push as “very dangerous.”
The Oligopoly Warning, Explained
Here’s the logic, stripped of jargon: when only a handful of companies can afford to comply with strict AI safety rules, those same rules quietly become a moat. Smaller competitors get squeezed out. Startups can’t keep up. The “safety-first” companies end up as the last ones standing — not by accident, but by design.
It’s a bit like a chess grandmaster proposing new rules that only grandmasters can follow, then calling it fair play for beginners.
Lonsdale isn’t alone in raising this concern. Regulatory capture — when an industry shapes the very rules meant to police it — is a well-documented pattern in fields ranging from banking to pharmaceuticals. Applying that lens to AI isn’t a stretch; it’s a reasonable question worth asking out loud.
“Safety Is Real” – But Fear Isn’t a Policy
To his credit, Lonsdale didn’t dismiss Artificial intelligence safety as nonsense. He drew a clear line between the risk itself and the response to that risk.
“Safety is a real thing,” he said, adding that regulators “have to be really careful how we respond” to fear-driven pressure.
That’s a fair distinction. A fire alarm is useful. Panic-buying every fire extinguisher in the store because someone yelled “smoke” is a different matter entirely.
Still Bullish on Anthropic
Despite the criticism, Lonsdale didn’t turn on Anthropic itself. He praised the company’s technical performance and stood by his investment.
He called Anthropic “the highest-performing company in the world over the last few years.” He added, “I’m proud to be an investor in Anthropic.”
That combination – criticizing a company’s lobbying strategy while praising its product – isn’t contradictory. It’s actually a pretty normal investor move. You can admire the engine and still question who’s steering the car.
A Chorus of Warnings From AI’s Biggest Names
Lonsdale’s comments land in a moment when several major tech leaders have publicly flagged the risks of advanced AI. Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and SpaceX CEO Elon Musk have each called for more caution around deploying increasingly powerful Artificial intelligence systems. Superpower Daily
That tension is at the heart of this story. The same executives sounding the alarm on AI risk are also, in Lonsdale’s view, shaping the policy response to that risk. Whether that’s responsible leadership or self-interested positioning depends entirely on who’s asked.
Probably both, if we’re being honest.
From AI Policy to the Battlefield
Lonsdale’s remarks didn’t stay confined to Artificial intelligence regulation. He also said warfare itself is changing, pointing to defense technology company Anduril, another firm backed by his investment portfolio.
He described deepening commercial ties between his portfolio companies and businesses in Ukraine, where the ongoing conflict has become a proving ground for new military technology. “We’re partnering very closely with them,” Lonsdale said, adding that some of his firm’s companies could eventually acquire Ukrainian businesses outright.
This isn’t happening in a vacuum. Ukraine’s battlefield has become a real-world testing ground for autonomous drones and AI-assisted targeting systems from multiple Western defense companies, a trend defense analysts have tracked since the war’s early stages. Lonsdale’s comments confirm that trend is deepening, not slowing down.
Who’s Accountable When AI Weapons Are Involved?
Autonomous weapons raise an uncomfortable question: when a machine helps make a battlefield decision, who answers for it?
Lonsdale’s answer was straightforward. He argued that responsibility should sit with the humans who built, approved, and deployed the system — not the system itself.
“There’s always a person responsible who is in charge of whatever system they studied, worked on, and approved,” he said.
He also pushed back on armchair judgment of wartime decisions made far from the front lines. “It’s very easy from the outside to judge these things when you’re not there in the fog of war,” Lonsdale said.
Fair point. Second-guessing a decision from a conference room is a very different exercise than making one under fire, with incomplete information and a clock running.
Is a Pentagon Job Next?
Lonsdale’s name has come up in discussions about a senior Pentagon role during an ongoing leadership shakeup at the Department of Defense. Asked directly, he kept his answer measured.
“My focus right now is on building my businesses and my community,” he said. However, he acknowledged he’d “consider” a direct request from a sitting president.
Translation: the door isn’t locked, but he’s not exactly waiting by it either.
The Bottom Line
Strip away the headlines, and Lonsdale’s argument boils down to a simple idea: good intentions and good policy outcomes aren’t automatically the same thing. A company can genuinely believe Artificial intelligence poses real risks and still end up writing rules that happen to protect its own market position. Both things can be true at once.
That’s not a reason to dismiss Artificial intelligence safety concerns. It’s a reason to ask who’s in the room when the rules get written – and who isn’t.
As AI regulation debates heat up in Washington and beyond, that question isn’t going away anytime soon.
