As users, we rely on digital spaces for information, while at the same time, these are increasingly being cluttered with “AI slop.” But what happens when the slop becomes convincing? If the past year in technology has made one thing clear, it is that artificial intelligence needs more guardrails. Leading artificial intelligence developer OpenAI has launched a second iteration of its artificial intelligence powered text-to-video generator, Sora.
Sora 2, launched Sept. 30, has sparked controversy over its ability to produce realistic videos using AI. All the platform requires is a simple text description to generate hyperrealistic footage. Within days of its launch, early adopters of the AI generation tool have used it for fun, creating humorous, fabricated videos of celebrities, skating animals or videos featuring popular fictional characters. While the old Sora could also generate high-quality videos, the new platform allows users to upload videos of real people and put them into AI-generated environments, even adding sound effects and dialogue. While some may just be looking for fun, others may use this new technology for more nefarious purposes.
With digital spaces already increasingly entrenched in disinformation, Sora 2 is adding fuel to the fire. Sora 2 and similar AI-powered video generating tools released by Meta and Google significantly lessen the weight of video proof. According to the New York Times, “the tech could represent the end of visual fact.” Sora 2 generated falsified videos of bombs, mass-shooting scares and fake security footage of crimes that never happened. Misinformation researchers say that lifelike scenes such as these can cause the truth to become blurry and create opportunities for fraud, bullying and intimidation. The platform gives us a glimpse into the near future, where it will be increasingly difficult to separate truth from fiction.
The launch of Sora 2 has also received a swift backlash from Hollywood studios and rights holders. The core of the dispute centers around who controls the likenesses of licensed characters and if intellectual property owners should be compensated. The platform initially launched with a controversial third-party rights model. There were some restrictions on copyrighted content, but rather than a more traditional opt-in model, rights holders were required to opt out of their works’ being used in Sora 2 on a case-by-case basis. Since its launch, after facing controversy, OpenAI has backpedaled on their original model and are shifting towards an opt-in approach and passing a cut of proceeds to rights holders.
OpenAI has made it clear with the company’s “media upload agreement” that in the case of any problematic content created using their technology, the users are to blame. Since the platform’s release, OpenAI has responded to backlash, shifting toward an opt-in model for rightholders and threatening to ban users who misuse media uploads. OpenAI’s main focus seems to have been ensuring the success of Sora 2 and only cleaning up the damage afterward. The lack of accountability from OpenAI highlights its prioritization of growth and profit over ethical safeguards.While Sora 2 does have some restrictions — it is still invite only, videos carry an AI label and some prohibited or copyrighted is blocked — the platform’s users have found workarounds. The launch of Sora 2 should serve as a warning to everyone and underscore just how easy AI has made it to produce false and misleading content that is convincing. Users should carefully observe how the controversy surrounding Sora 2 plays out, as it will likely set a precedent for how AI tools will be handled in the future.
