OpenAI CTO Admits Sora's Training Data Sources, Sparking Controversy
-
OpenAI's latest achievement, Sora, has revealed a glimpse to the public, sparking a controversy over the sources of its training data. In a recent interview, OpenAI's Chief Technology Officer, Mira Murati, disclosed that Sora was trained on publicly available and licensed data. However, when asked whether YouTube, Facebook, or Instagram video data was used, she expressed uncertainty. This statement has drawn widespread skepticism, with many criticizing Murati for not being transparent enough, and some even accusing her of lying.
Despite the controversy, the launch of Sora remains highly anticipated. This new AI model can generate hyper-realistic scene videos based on text prompts, marking a significant breakthrough in text-to-video technology. However, although Sora's presence is ubiquitous on media platforms, it will not be made available to the public in the short term. This is because the videos generated by Sora are so close to reality that they may trigger a series of issues, requiring further improvements and time for people to adapt. To make Sora more realistic, OpenAI plans to incorporate audio generation capabilities and continuously optimize the model, including improvements in coherence, usability, and cost. Additionally, they intend to introduce user editing features, allowing users to further refine Sora-generated videos for better results and more accurate content expression.
However, beyond technical exploration, the launch of Sora has raised a series of questions and concerns. One major issue is the source of training data, which directly relates to copyright concerns. Although OpenAI claims to use publicly available and licensed data, the authenticity of this claim has been questioned, sparking public concern and dissatisfaction.
In response to the controversy, OpenAI needs to address public doubts with greater transparency and honesty. At the same time, they must take measures to ensure that content generated by Sora does not involve illegal or harmful material, safeguarding public interests and safety. In the AI era, technological advancement must align with societal oversight and regulation to ensure long-term sustainability and stability.