Marques Brownlee, popularly known as MKBHD, highlights how Google's Veo 2 AI video generator, owned by Alphabet Inc. (NASDAQ:GOOG) (NASDAQ:GOOGL), outshines OpenAI's Sora model.
What Happened: On Thursday, Brownlee shared a YouTube Short of his observations pointing out that while Sora often displayed its AI nature through inaccuracies in movement and physics, Veo 2 delivered more realistic results.
However, he acknowledged that Veo 2 still has flaws, such as issues with complex scenes and garbled text.
One of the standout points in Brownlee's comparison was his emphasis on YouTube data—something that sets Veo 2 apart from other models, including OpenAI's Sora.
He suggested that the difference between the two is like "using a little bit of YouTube data" versus "owning YouTube and just using all of it."
In March this year, former OpenAI CTO Mira Murati's uncertainty about whether YouTube videos were used to train Sora created controversy. The following month, YouTube CEO Neal Mohan said that using YouTube videos in such manner would violate their policy.
When asked if Google is using YouTube to train Gemini, Mohan said, "Google uses YouTube content really in accordance. Again, back with those terms of service or individual contracts that we might have with creators or uploaders to our platform."
Subscribe to the Benzinga Tech Trends newsletter to get all the latest tech developments delivered to your inbox.
Why It Matters: Earlier this week, Elon Musk, who co-founded OpenAI in 2015 and left the AI startup in 2018 over some differences, expressed admiration for Google's Veo 2.
On the flip side, Marc Andreessen has praised OpenAI's Sora model, suggesting it may have solved a significant challenge in robotics.
Earlier this year, Amazon.com Inc. (NASDAQ:AMZN) also introduced its AI-powered video tool for advertisers, joining Adobe's Inc. (NASDAQ:ADBE) Firefly Video Model in the competitive market.
In October 2024, Meta Platforms Inc. (NASDAQ:META) unveiled Movie Gen, a new tool that generates videos, images, and audio based on text prompts.