Sora AI: The Beginning of the End?

Over the last few years, AI-generated images and videos have advanced at a pace that has surprised even those who once dismissed the technology as a short-lived novelty. Sora, OpenAI’s newest video model, represents the most dramatic leap yet. Its emergence raises questions and concerns about labor, intellectual property, and whether we’ll lose the ability to distinguish synthetic content from human-made labor.

Sora is an AI model designed to generate both animated and live-action video from text descriptions. Sora is capable of producing videos that strikingly resemble pre-existing intellectual properties (IPs) such as “Family Guy,” “South Park,” and “Spongebob Squarepants.” Although OpenAI has made attempts to restrict the generation of copyrighted material, users have discovered loopholes through carefully made prompts. By describing a show’s visual style, character appearance, and personality – without directly stating names or the IP directly – users can still coax the AI into generating copyrighted material.

The implications for creative labor are even more serious. Animation is produced by entire teams of people, artists, animators, editors, sound designers, etc all have a hand in production, Sora’s ability to automate large chunks of this work pipeline means that these workers could face reduced demand or complete replacement. It may only be a matter of time before many creative jobs become obsolete.

Outside of creative material, Sora AI is also capable of generating highly realistic live-action footage. On platforms such as Facebook, Instagram, and Tiktok, AI-generated videos have already flooded the short-form content space. Granted, some of these clips are intentionally absurd, designed to be obviously fake and consumed with a sense of irony; however, increasingly, AI-generated content is being produced with engagement in mind, crafted to look real and monetized without viewers realizing they’re watching synthetic content.

This problem extends to long-form content as well. Some YouTube channels are made up entirely of synthetic content. Lots transcribe someone else’s video – usually a video essay made by a real person – feed the script into AI, then generate the visuals, rehashing the entire video with a fresh coat of AI paint. These channels actively siphon attention directly away from the original creators.

This shift has raised concerns about the indistinguishability of AI-generated content, whether it be real footage, animation, or anything in between may soon become, and how easily misinformation can spread when realism is no longer a reliable metric. While the threat may not be Sora AI itself, it is a major step in the right direction for AI generated video.