A director created a deepfake of Sam Altman, and he was unexpectedly hooked

Director: Adam Bhalla Lough was not prepared to produce a documentary about a digital simulation of Sam Altman.

But after nearly 100 days of texting and emailing OpenAI’s CEO for an interview — with no response, he claims, and with funders hounding him for success with his original pitch — Lough was at his wits’ end.

I’ve exhausted almost every angle. “Once I got to that point, I gave up and focused on gate-crashing OpenAI,” he says. Although he used a similar tactic in his 2023 Emmy-nominated documentary Telemarketers– A record of industry-wide corruption in telemarketing – a style of filmmaking he was not comfortable with. “It was a fortress. I managed to sneak through the gate, and immediately security grabbed me and physically removed me from the building.”

This is how it begins Deepfakes Sam AltmanLove’s portrait of how artificial intelligence is reshaping society and his quest to speak to the man behind it. When his original plan failed, he found inspiration from Altman himself. “The Scarlett Johansson controversy has erupted,” he says. In 2024, the actress publicly criticized OpenAI because it appeared to copy her voice for its new AI voice assistant Sky. “At that point I had the idea of ​​doing a deepfake.” (In a May 2024 statement, Altman apologized to Johansson and said Skye’s voice was “never intended to sound like hers.”)

What originally starts out as simple sound reproduction balloons turns into a full-fledged Altman deepfake called the Sam Bot, which Lough travels to India to create. This being Lough’s film, nothing goes according to plan. Without spoiling too much, Sam Bot eventually becomes its own entity, and the film takes on an even stranger and more suggestive journey from there. “There are similarities between this film and… Terminator 2: Judgment Day“But there is no violence,” he says. Love grew up during what he calls the “AI 1.0 era.” His obsession with James Cameron Terminator 2 He had a huge influence on his craft.

Deepfakes Sam Altmanwhich is based in part on a New York magazine story that casts Sam Altman as a modern-day Oppenheimer, features commentary from former OpenAI safety engineer Heidi Khalaf, who told Love: “We’re starting to see OpenAI dipping its toes into military uses, and I can’t imagine using something like Dall-E and ChatGPT for military assistance. That really scares me, given how imprecise these systems are.”