AI
YouTube star Marques Brownlee has pointed questions for OpenAI after its Sora video model created a plant just like his
Blake Dodge
2024-12-10T19:23:57Z
Marques Brownlee's Sora review.
Marques Brownlee's Sora review. Marques Brownlee
On Monday, OpenAI released Sora, an AI video-generator, with hopes of helping creators.
One such creative, YouTuber Marques Brownlee, wants to know if his videos were used to train Sora.
"We don't know if it's too late to opt out," Brownlee said in his Monday review of Sora.
On Monday, OpenAI released its video generator Sora to the public.
CEO Sam Altman showed off Sora's capabilities as part of "Shipmas," OpenAI's term for the 12 days of product launches and demos it's doing ahead of the holidays. The AI tool still has some quirks, but it can make up to 20-second videos from a few words of instruction.
During the launch, Altman pitched Sora as an assistant for creators and said that helping them is important to OpenAI.
"There's a new kind of co-creative dynamic that we're seeing emerge between early testers that we think points to something interesting about AI creative tools and how people will use them," he said.
One such early tester was YouTuber Marques Brownlee, whose tech reviews have garnered roughly 20 million subscribers. One could say this is the kind of creator that OpenAI envisions "empowering," to borrow execs' term from the livestream.
But as part of his Sora review, Brownlee didn't sugarcoat his skepticism, especially when it comes to how the model was trained. Were his own videos used without his knowledge?
This is an open mystery, and a controversial one. OpenAI hasn't said much about how Sora is trained, though experts believe the startup downloaded vast quantities of YouTube videos as a key part of the model's training data. There are no legal precedents for this practice, but for Brownlee, the lack of transparency is sketchy, he said.
"We don't know if it's too late to opt out," Brownlee said on Monday.
In an email, an OpenAI spokesperson said Sora was trained using proprietary stock footage and videos available in the public domain, without commenting on BI's specific questions.
In a blog post about some of Sora's technical development, OpenAI said the model was partly trained on "publicly available data, mostly collected from industry-standard machine learning datasets and web crawls."
Brownlee's big questions for OpenAI
Brownlee threw dozens of prompts at Sora, asking it to generate videos of pretty much anything he could think of, including a "tech reviewer talking about a smartphone, while sitting at a desk in front of 2 displays."
Down to the man's hand gestures, Sora's rendering was believable. But Brownlee noticed something curious. Sora added a small fake plant in the video, eerily matching his own fake plant in the real world.
Marques Brownlee's Sora review. Marques Brownlee
Throughout the review, the YouTuber showed all manner of "horrifying and inspiring" results from Sora, but this one seemed to stick with him the most. The plant is generic-looking, to be sure, but for Brownlee, it's a reminder of the unknown behind these tools. The models don't create anything fundamentally novel; they are merely predicting frame after frame based on patterns that they've recognized in "source" or original material.
Related stories
"Are my videos in that source material? Is this exact plant part of the source material? Is it just a coincidence?" Brownlee said. "I don't know." BI asked OpenAI about these questions specifically. The startup didn't address itn its comments.
Marques Brownlee's Sora review. Marques Brownlee
Brownlee discussed Sora's guardrails at some length. One feature, for example, can make videos from images that people upload, but it's pretty picky about weeding out copyrighted content.
Thousands of people are already commenting on Brownlee's post. A few of them thought it was ironic that Sora was careful to steer clear of intellectual property — except when it came, supposedly, to the millions of people whose work was used to produce it.
Somehow, their rights don't matter one bit, one commenter said, "but uploading a Mickeymouse? You crook!"
In an email to BI, Brownlee said that he was looking forward to seeing the conversation evolve.
Millions of people. All at once.
Overall, the YouTuber gave Sora a mixed review.
Outside of its inspiring features — it could help creatives find fresh starting points, at a minimum — Brownlee feared that Sora was a lot for humanity to digest right now.
The model does a good job of refusing to depict dangerous acts or use images of people without their consent, Brownlee said. Though it's easy to crop out, it also adds a water mark to the content it makes.
Sora's relative weaknesses might provide another layer of protection from misuse. In Brownlee's testing, the system struggled with real-life depictions of object-permanence and physics. Objects will pass through each other or disappear. Things might seem too slow, then suddenly too fast. Until the tech improves, at least, this will help people spot the difference between, for example, real and fake security footage.
But from here, the videos are only going to get better, per Brownlee.
"The craziest part of all of this is the fact that this tool, Sora, is going to be available to the public," he said. "To millions of people. All at once."
"And it's still an extremely powerful tool that directly moves us further into the era of not being able to believe anything you see online," he added.