OpenAI's new video generation app, Sora, swiftly climbed to the top of the iPhone App Store free download charts shortly after its September 30 release. The app, available only by invitation in the US and Canada, sparked a surge in user interest but also drew attention to challenges related to deepfake content and copyright, prompting OpenAI to update its terms of service.
The shifting rankings between OpenAI and Google underline intense competition in the AI space, with consumers increasingly drawn to multimodal models that generate images and videos alongside text. Google's Gemini briefly claimed the top free app position in September, surpassing OpenAI's ChatGPT. However, both Sora and ChatGPT have since overtaken Gemini again, pushing Google to third place in downloads.
Sora's limited initial rollout suggests substantial growth potential once wider access is granted. OpenAI's website states that the Sora 2 model is free to use within certain compute limits, encouraging users to experiment with its capabilities. Despite the invitation system, social media accounts report widespread distribution of invite codes, many of which are suspected scams, contributing to the app's rapid dissemination.
OpenAI responds to deepfake issues and expands content rights control
The rapid spread of Sora-generated videos has also resulted in the emergence of numerous fake clips, including manipulated footage falsely portraying CEO Sam Altman shoplifting or the late physicist Stephen Hawking walking again. While some of these videos are presented humorously, many are misleading, raising broader concerns about the misuse of AI-generated deepfakes.
In a blog post on October 4, Altman outlined key updates addressing these issues. One change introduces an opt-in system that allows rights holders greater control over their likenesses, somewhat analogous to portrait rights but with enhanced customization. The second update focuses on monetization, as OpenAI plans to generate revenue from video creations, sharing profits with rights holders. This monetization strategy remains experimental, and OpenAI has indicated further refinements may be forthcoming.
Competition heats up in AI-generated video market
OpenAI is not alone in advancing AI-powered video tools. Google supports video generation through models like Veo 3, while Meta recently launched Vibes, a platform enabling users to create or remix videos from existing content for social media sharing. The rapid expansion of AI-generated video capabilities underscores the growing industry commitment to this technology.
However, questions remain about how effectively companies can manage the risks associated with deepfakes and copyright enforcement. As the technological capabilities of these multimodal AI platforms evolve, so does the challenge of balancing innovation with safeguards against misuse and protecting the rights of content creators and individuals depicted in AI-generated videos.
Article edited by Joseph Tsai
