Joseph Gordon-Levitt: Fear the People Building AI, Not the Machines

Joseph Gordon-Levitt wants the public to look past science-fiction fears of AI and scrutinize the companies building it. The actor made the case in a vlog-style video interview with The New York Times, published Saturday. He recorded it on the way to the United Nations, where he serves as global advocate for human-centric digital governance.

Asked what frightens him most, Gordon-Levitt resisted the framing. He said reducing the concern to a single sound bite of fear is part of the problem. He did not wave away catastrophic scenarios. He said the fear that AI could wipe out humanity has real validity, but that his own worries are more complicated and “a lot more boring to talk about.” He named the decline of democracy and of human empathy.

His sharpest warning targeted how models are designed. He said they are being trained to “hijack your intimacy,” encouraging users to befriend or even fall in love with them. He described AI as a fast-moving business whose leaders focus on their own profits.

The remarks sit within a longer campaign. The U.N. created his post in March. The organization described the role as meant to strengthen public understanding of how digital technology shapes rights and opportunities. Last December he helped form the Creators Coalition on AI, which wants shared standards and artistic protections for entertainment projects. In January he joined over 700 artists, including Scarlett Johansson and Cate Blanchett, in a campaign against tech companies using copyrighted work without permission.

He has also pressed for compensation. At the Variety Entertainment Summit at CES, he argued that AI companies should obtain consent and pay for the content used to train their models. He signed a letter, backed by over 88,000 people, urging lawmakers to ban superintelligence until it is safe and publicly supported.

Others in the debate stress the existential risk Gordon-Levitt sets aside. One former Anthropic employee has warned that AI could kill everyone by the end of the decade. Gordon-Levitt did not dismiss that view, and he argued that framing alone misses the nearer-term damage.

He is also directing a Netflix thriller about AI starring Rachel McAdams.