Sora, an AI application that creates videos from text or photos, recently demonstrated by OpenAI, has attracted attention in the online community because of its high realism and easy operation... but also raises many concerns.
Accordingly, Sora can create videos up to one minute long, with multiple characters, various types of motion in both the foreground and background, from just text commands, or from still images.
One of the most shared videos on social media shows a girl walking down a brightly lit street in Tokyo. The video is created from the following descriptions: "She is wearing a black leather jacket, a long red skirt, black boots, and a black purse. She is wearing sunglasses and red lipstick. She walks confidently and calmly. The street is wet and reflective, creating a multi-colored backlight effect. Many people are walking."
With this video, many people believe that the shared videos show very realistic details and images created by AI, making it difficult for users to distinguish which videos are created by AI. However, if you observe closely the movements of the arms, legs, etc., you can see that the images are from AI. Currently, OpenAI has not released any trial versions for users other than the 10 sample clips available on the page https://openai.com/sora.
Sora also raises concerns in the context of Deepfake spreading on the Internet, especially fake videos that can have negative impacts on society. Therefore, OpenAI is building a classifier that can identify video clips as Sora products to help identify AI-generated content.
BA TAN
Source
Comment (0)