OpenAI, the creators of ChatGPT, have introduced their latest AI model capable of producing lifelike one-minute-long videos solely from textual inputs.The creators of ChatGPT, OpenAI, have released an impressive new AI model. Named Sora, this model has the capability to generate complete one-minute videos based solely on text prompts. According to the OpenAI Sora blog, the aim is to train AI to comprehend and replicate real-world motion, thus assisting people in solving problems that necessitate interactions with the physical environment.
OpenAI asserts that Sora possesses the ability to craft elaborate scenes featuring multiple characters, precise movements, and detailed backgrounds. The model not only grasps the user’s prompts but also interprets how these elements translate into real-world scenarios. According to OpenAI, “The model demonstrates a profound understanding of language, allowing it to accurately interpret prompts and generate compelling characters capable of expressing vibrant emotions. Additionally, Sora can generate multiple shots within a single video, maintaining consistency in characters and visual style.”
Despite generating significant excitement on the internet, the OpenAI Sora model has raised valid concerns, as noted by popular YouTuber Marques Brownlee, also known as MKBHD. In a post, he remarked, “Every single one of these videos is AI-generated, and if this doesn’t concern you at least a little bit, nothing will.”
However, OpenAI is not avoiding the discussion on safety. The company states that before integrating Sora into OpenAI’s products, they will implement several essential safety measures. This includes collaborating with red teamers, experts in fields such as misinformation, hateful content, and bias, who will extensively test the model to identify potential vulnerabilities. Moreover, OpenAI will develop tools to detect misleading content, such as a detection classifier capable of identifying videos generated by Sora.
OpenAI will also utilize existing safety protocols developed for its products that utilize DALL·E 3, which are applicable to Sora as well. For instance, within an OpenAI product, OpenAI’s text classifier will scrutinize and reject input prompts that violate usage policies, like those requesting extreme violence, sexual content, hateful imagery, or celebrity likeness. The company has also established robust image classifiers to review every frame of generated videos, ensuring compliance with usage policies before user access.
Additionally, OpenAI states that it actively engages policymakers, educators, and artists worldwide to address concerns and explore positive applications of this new technology. OpenAI expresses, “We’ll engage policymakers, educators, and artists worldwide to understand their concerns and identify positive use cases for this new technology. Despite extensive research and testing, we cannot predict all the beneficial ways people will use our technology, nor all the ways people will abuse it. That’s why we believe that learning from real-world use is a critical component of creating and releasing increasingly safe AI systems over time.”
Presently, Sora is solely available for red teamers to examine critical areas for potential issues or risks. OpenAI is also granting access to visual artists, designers, and filmmakers to gather input on how to enhance the model. The blog states, “We’re sharing our research progress early to start working with and getting feedback from people outside of OpenAI and to give the public a sense of what AI capabilities are on the horizon.”