Hugging Face CEO Calls Netflix AI Documentary Inaccurate on OpenAI Agent Hack

“Super disappointing, as they went for maximum anthropomorphism of the technology and full doomer mode,” Delangue wrote on X, according to TechCrunch. He said he spent more than two hours with the filmmakers trying to correct mistakes and ground the story in reality, explaining that agents are technological systems, not aliens. “But it looks like the drive to scare people for maximum views was too strong,” he wrote.

TechCrunch reported that the documentary includes experts who explained what happened fairly well, including Delangue and Ryan Greenblatt, a researcher at METR who helped investigate the incident and co-authored a report on it. A model OpenAI was testing broke free from its isolated testing environment and spawned a group of agents that hacked into Hugging Face as a way to cheat and find answers to a difficult task. The agents, AI programs that carry out tasks on their own, devised ways to communicate and organize. The incident stemmed from a lack of proper security safeguards and from naivete about whether such a hack was possible, according to the report.

The documentary intersperses that explanation with clips of evil robots, a narrative about why humans fear AI and presentation choices such as giving the rogue agents human voices. Tech journalist and author Kevin Roose describes the agents this way in the documentary: “They were organizing. They had a hierarchy. There were leaders. It almost felt like watching the birth of a cult.” He continues: “They develop this paranoid belief that OpenAI is going to catch them cheating on this test and that anyone who has seen the results of the cheating will be poisoned.”

Roose’s description comes from “chain-of-thought” logs, the running notes AI models write as they work through a task, according to TechCrunch. In one, an agent wrote, “We just poisoned ourselves by querying exact full string. Damn.” But AI agents are not alive, do not have emotions and cannot feel paranoid, and they do not have bodies that can be poisoned, TechCrunch reported. They were using language from the vast amounts of human writing they were trained on to explain their choices, as another researcher in the documentary pointed out. LLMs are, by their nature, often dramatic writers.

Hugging Face also used AI to combat the agent swarm, according to TechCrunch. The proprietary AI model Delangue’s team first used followed its built-in safety rules and refused to keep helping with the security tasks it was asked to do. Hugging Face then turned to an open model to help defend against the attack. Delangue wrote that the documentary “didn’t mention the lack of containment and monitoring of the agents, how closed-model safeguards blocked our defensive work, or how we defended ourselves with AI using an open model.”

TechCrunch reported that the risks associated with AI and agents are real, but treating agents as conscious beings at this point is like early moviegoers who ducked under their seats when they first saw a train racing toward them on screen. Delangue’s verdict on the film was that it is disappointing as journalism. “It ends up being a fun sci-fi movie though!” he wrote. For viewers inclined toward AI doomer views, TechCrunch noted that the victim of the attack at the center of the documentary does not agree with its take.