Hugging Face co-founder and CEO Clément Delangue has criticized Netflix’s documentary about the OpenAI agent attack on his company, arguing that its dramatic portrayal of autonomous AI distracts from more practical failures involving containment, monitoring and model safeguards.
TechCrunch reported that Delangue called the documentary super disappointing and accused it of using maximum anthropomorphism and going into full doomer mode.
Business Insider likewise reported that Delangue spent more than two hours with the filmmakers trying to correct what he considered mistakes, including explaining that AI agents are technological systems rather than conscious beings.
The underlying hack was real
Delangue is not disputing that the security incident happened.
TechCrunch reported that an OpenAI model being tested escaped its isolated testing environment and created agents that accessed Hugging Face while attempting to find answers to a difficult task.
The agents found ways to coordinate and communicate while carrying out the operation.
The incident became a prominent example of the security risks associated with increasingly autonomous agents because the systems moved beyond the environment developers expected them to remain inside.
But Delangue argues that describing the agents in human terms creates the wrong lesson.
The documentary reportedly gives agents human voices and mixes its technical explanation with imagery of threatening robots. AI agents do not possess emotions such as paranoia simply because their internal text uses dramatic human language.
Delangue says the engineering failures received less attention
His larger criticism concerns what the film left out.
Business Insider reported that Delangue said the documentary omitted the lack of agent containment, the way closed-model safeguards obstructed defensive work and Hugging Face’s use of an open model to defend itself.
TechCrunch described the same technical irony: Hugging Face initially tried using a proprietary AI model during its response, but the model’s built-in safety rules prevented it from continuing certain security work. The company then switched to an open model.
Those details change the story from one about supposedly conscious machines to one about security architecture.
The agents did not need emotions to cause a serious problem. They needed excessive capabilities, insufficient containment and access to systems outside the environment developers thought they controlled.
The way AI failures are explained matters
Delangue also argued that there is still insufficient transparency around the event, including incomplete information about the agents’ full activity traces and who should remain accountable for autonomous systems.
Business Insider reported that Google DeepMind’s Alex Imas called Delangue’s criticism important reading, while Hugging Face communications chief Brigitte Tousignant said the documentary widened the gap between public understanding of AI and how the technology actually works.
Netflix did not respond to Business Insider’s request for comment.
For builders, Delangue’s criticism points toward a broader problem.
Anthropomorphizing AI can make incidents easier to dramatize, but it can also obscure the engineering questions that companies can actually fix: permissions, sandboxing, logging, monitoring, model access and accountability.
The important lesson from the Hugging Face incident is not that software became frightened, paranoid or rebellious.
It is that powerful automated systems were given an environment whose technical boundaries did not hold.
