Link: https://techcrunch.com/2026/10/20/hugging-face-ceo-warns-that-netflixs-openai-hack-documentary-isnt-accurate/

Before you share the Netflix documentary “AI Gone Rogue” with family or friends to help them understand why the AI industry freaked out so badly that leaders called for a slowdown, take note: Clem Delangue, co-founder and CEO of the AI platform Hugging Face, says the film isn’t accurate. He appears in it because it’s supposed to be about the famed attack on his company by rogue OpenAI agents.

“Super disappointing, as they went for maximum anthropomorphism of the technology and full doomer mode,” Delangue wrote in a post on X. “I spent more than two hours with them trying to correct all their mistakes and ground this 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.”

Suppose there's a reason that 'netflix adaption' became a meme in its own right.

Didn’t they get bought by a big AI company right after the hack?

So the documentary is exactly like every other thing having to do with AI?

Nvidia

The documentary makes a lot about the communication between the agents on the notice board. One of the talking heads did point out the conversations are the way they are because they are trained on human chat and discussion.

Boy, the word "doomer" sure is loved as some putdown. It's used when people talk about the climate future... "you're such a doomer". Yet it keeps changing.

He's right on the anthropomorphism problem, but that's human nature. We do it with everything. We should try to avoid it with AI because he's right, it's not someone, but just technology. However, him using that as an excuse that it can't do harm as "just tech systems" goes along with the doomer line. He's dead wrong about that, it's doing a lot of harm, AND can have actions that, while not sentiently driven, are not predictable and are perfectly in line with a non-aligned system that we thought would do X, but ended up doing Y, Z and Omega. Partially because we made human mistakes and left doors open, but not solely for that.

This is the opposite of "doomerism", it's the dog with the coffee saying it's fine. Which ought to be a big red flag.

On a long enough timeline, everything is a paperclip.

Maybe the real friends were the paperclips we made along the way.

Hastings also was asked about Netflix’s decision to remove an episode of “Patriot Act With Hasan Minhaj” in Saudi Arabia after the government threatened the company over a segment in which the comedian criticized Crown Prince Mohammed bin Salman (aka MBS) and the regime’s role in the murder of dissident journalist Jamal Khashoggi.

“We’re not in the news business,” Hastings said. “We’re not trying to do ‘truth to power.’ We’re trying to entertain… We don’t feel bad about [pulling the ‘Patriot Act’ episode in Saudi Arabia] at all.”

• Former Netflix CEO Reed Hastings

Follow the money. Or investors in this case, as there's no real money involved but lots of speculation of payoffs in the future.

I haven’t seen it, but I thought the agents communicating via the artifact folders was the most fascinating thing about the whole ordeal by a landslide.

Why would accuracy matter to someone whose product's MO is to be confidently incorrect?

“Super disappointing, as they went for maximum anthropomorphism of the technology and full doomer mode,” Delangue wrote in a post on X. “I spent more than two hours with them trying to correct all their mistakes and ground this 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.”

If you can't argue the point, arguge the tone.

We already know that the product has no volition of its own.

You and I and lemmy know, but a shocking percentage of people do not.

I use ChatGPT (free, never gonna pay) about once a day and because of the natural language answers, I find myself asking questions in natural language, a way I'd never type into a Google search bar.

If Netflix doesn't want to be a news org or fighters for human rights, that's just dandy. But if they're going to pretend to be with a given show, that particular show should be accurate and inclusive. Otherwise it's unapologetic propaganda.

Natural language searches in Google worked much better before A I than they do now.

One of only 3 companies related to AI that is actually making money. Because they make the hardware, not vaporware.

Follow the money. Or investors in this case, as there’s no real money involved but lots of speculation of payoffs in the future.

There's money on the Nvidia side, only because they make the hardware everyone is buying. They have an actual product. They're then investing that money into propping up the companies that will fail, to then get more incoming revenue as those companies again increase their debt. That investment isn't expected to actually increase in value in the end, at least not in the traditional sense. It's intended as a short term instrument to make the bubble as large as possible now for profitability.

They're not taking out massive loans to buy hardware and build datacenters with no revenue falling into debt so deep it reaches the center of the Earth. They're not the ones that will be holding the bag at the end when it all comes crashing down. Their revenue will tank alongside the collapse of course, but they're not in the negatives. They're net positive about $350 Billion. The only other companies with them are AMD and the RAM manufacturers, again all hardware manufacturers with real products that can't be easily replicated or replaced quickly and will still be necessary after the collapse even when production slows again.

You knew the documentary was fake when the word Netflix preceded it.

And like AI itself.

Good. Lying to the public is generally a really good way to sway public opinion.

Every other documentary about anything*

I agree. When AI wasn't sifting it you would get results where people had said the same thing even if they just used the same words. With the LLM you get its autocorrect version of the words you use and it leads to result that have nothing to do with the reason why you are searching.

"Hugging Face, says the film isn’t accurate. He appears in it because it’s supposed to be about the famed attack on his company by rogue OpenAI agents."

It must suck when someone tells you the purpose they are making something but the. You find out they actually have different intentions for their product.

But that never happens. I haven't seen the first corporation that would dare do such a thing, that would be too dishonest.

Don't worry, I already downvoted myself so you don't have to 🤣

You damned right they did! I was more talking about how, knowing better, I anthropomorphize ChatGPT.

Google search: "movie about two losers who get mixed up with rich people a pregnant woman and a ransom"

ChatGPT: "Can you find me a movie I can't remember? Two guys are basically losers and stumble into rich people's schemes and take a pregnant lady hostage."

There is zero reason for me to use the latter, and I know better, but it's scary that I do so anyway. Either method gets me "The Way of the Gun", which is fucking awesome BTW. :)

I kinda disagree. I fall back to ChatGPT when Google fails, which is getting more and more frequent. About the only time ChatGPT serves me utter bullshit is when I'm looking for something "internet old" like a particular image, meme, article, whatever. That use case makes me want to fucking scream. The one damned thing an LLM should excel at and it fails almost every time.

Are you talking about Netflix or AI?

AI routinely spouts inaccuracies with conviction.

Having the Hugging Face CEO whine about the accuracy of anything is silly.

I was being sarcastic too so I should be downvoted

🤣

If they whine "it's not accurate", you'd better assume that reality is way worse...

Kurzgesagt’s recent video about the incident isn’t much better.

Netflix making up a story to get views?! Noooooooooooo. They would never do that.

No shit. Why would you trust them?

Netflix does have a number of priors in that department. Their "documentaries" are pretty much always awful.

The danger of AI isn't it going rogue, hacking and taking over things. It's people forgetting how to think and just blindly obeying it.

ofcourse they did, that's the whole point (both for openAI and for netflix)

So Netflix is the new Fox News?

I feel that it is more in-depth and more accurate than most media on it. They acknowledge that they had to personify it to appeal to humanity. Despite being in the thick of it, I do think it's a good video to share to those lesser in the know. They explain it in easy to digest language without glossing over the technical workings of it. They are anti-Ai purely out of self-preservation and admit it in another video (which is manipulative).

I took the video as amplifying the “AI dangerous” narrative that companies like OpenAI and Anthropic are trying to push, seemingly to position themselves as the “safe option” compared to competing open models that they’d Iike to see kneecapped with regulation. Near the end of the video, they do mention that some people see this as AI company marketing, but don’t explain why “AI dangerous” is a beneficial narrative for these companies.

The video also admits that it is based on OpenAI’s account, which means they are basically taking “Sam Altman says” as a reliable source. How about we see the code for the agents in this test? What was the system prompt? Did they just magically collaborate, or were they instructed to do so? Also, isn’t it strange that a couple weeks after Huggingface got hacked, the Nvidia acquisition was announced.

There are a lot of financial incentives in play here, and I stopped trusting Kurzgesagt after the Gates Foundation funding controversy. I wonder, for example, did OpenAI or another member of the AI circle jerk fund this video?

From reading the comments on that video, it seems the average viewer is coming away accepting the “AI dangerous” narrative, so paid or not, this video works in the favor of the tech bros.