OpenAI has launched Sora 2 and a new social media app called Sora, marking its boldest move yet into the world of AI-generated content and digital social interaction. The app, currently invite-only and limited to users in the U.S. and Canada, functions like a TikTok alternative, where all videos are generated by AI in response to text prompts. Each video is capped at 10 seconds and can be created using OpenAI’s latest model, Sora 2, which offers improved realism, better physics simulation, and the ability to generate synchronized audio, including dialogue, sound effects, and background music. Despite its technical advancements, Sora has quickly become a hotspot for copyrighted material. Users have generated videos featuring iconic characters like Sonic the Hedgehog, Pikachu, Solid Snake, and Darth Vader—many of whom are protected by intellectual property laws. Some videos mimic the visual style and dialogue of popular shows like Rick and Morty and Wednesday, even replicating specific fonts and character likenesses. In one striking example, a user prompted Sora to generate a scene in Studio Ghibli style and it produced a clear homage to The NeverEnding Story. These outputs suggest that Sora 2 was trained on vast amounts of copyrighted media, including films, TV shows, and video games, without explicit permission from rights holders. OpenAI’s approach to copyright is controversial: it assumes permission unless a rights holder explicitly opts out. Disney and other studios have already requested to be excluded, but that doesn’t stop the model from generating content based on its training data. The company claims it is “using it unless we’re explicitly told not to,” a strategy that could lead to widespread infringement claims. The app’s “Cameos” feature allows users to upload a short video and audio clip to create a digital avatar that can be inserted into any AI-generated scene. This feature has led to viral videos of public figures like Sam Altman—whose likeness is available for use—appearing in absurd or humorous scenarios, such as stealing GPUs from Target or being turned into a skibidi toilet. Other users have inserted their own likenesses into NFL games, raising questions about the use of logos and trademarks. While OpenAI touts consent-based controls—users can revoke access, delete videos, and view all uses of their likeness—the system has flaws. Exported videos can be shared without watermarks, and metadata can be stripped. This raises concerns about deepfakes, identity misuse, and reputational harm. Despite its technical prowess, Sora’s social appeal remains uncertain. While users are entertained by novelty and creativity, the lack of authenticity and human connection may limit long-term engagement. Unlike real-life content, AI-generated videos lack emotional resonance, and the constant awareness that they are artificial could diminish their impact. OpenAI’s real goal may not be to replace TikTok, but to build a data-rich ecosystem that fuels model improvement and establishes a dominant position in AI-native social media. The company is betting that users will keep creating and sharing content, feeding a self-reinforcing loop of innovation and engagement. For now, Sora represents both a technological leap and a legal and ethical minefield. As AI-generated content floods social platforms, the tension between innovation and copyright protection will only intensify. The real test will be whether people stay engaged beyond the novelty—and whether the law catches up.