Skip to content
Advertisement
Palmer Luckey tackles AI control and accountability at DealBook Summit

Palmer Luckey tackles AI control and accountability at DealBook Summit

The Anduril founder argued that humans, and specifically elected governments, must stay accountable for how advanced AI gets deployed

Can people actually control the AI models they are building? Palmer Luckey, the founder of defense tech company Anduril Industries, took on that question at the New York Times DealBook Summit.

His appearance came on November 18, 2026, at an event where AI ethics and policy dominated the agenda. Luckey’s remarks centered on AI controllability and emergency preparedness, two topics that sound abstract until something goes wrong.

What Luckey argued

The core of Luckey’s position is that AI should make decisions better in high-stakes situations, not worse. He paired that with a firm insistence on human accountability for how advanced technology gets deployed.

Luckey has long backed the use of autonomous AI weapons when they reduce civilian casualties. His argument is blunt: there is no moral high ground in choosing inferior technology.

Advertisement

He also draws a hard line on who gets to decide. In his view, decisions about deploying AI weapons belong to elected governments, not private companies.

His stated reason is to prevent corporatocracy, a system where corporations effectively hold the levers of state power.

Who Luckey is and why his view carries weight

Luckey’s path into defense was not the obvious one. He co-founded Oculus VR, the virtual reality company that Facebook acquired in 2014.

He went on to found Anduril Industries in 2017. The move took him from consumer headsets to military technology.

The DealBook Summit stage put him alongside other major voices in AI, including OpenAI’s Sam Altman. The broader conversation at the event focused on AI’s ethical dimensions and its expanding role in business and public policy.

The control question, unpacked

When people ask whether humans can control AI, they usually mean one of two things. Some worry about technical control: whether a system will reliably do what it was designed to do.

Others worry about institutional control: who sets the rules, who approves deployment, and who is liable when something fails. Luckey’s emphasis leans heavily toward the second category.

His argument suggests that technical safeguards alone are not enough. Without clear lines of accountability, even a well-behaved system can be used in ways nobody voted for.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.