Sora 2 and the Future of AI Video: Innovation, Copyright, and Disinformation

1. Xin, Cheng. "Sora 2 app icon displayed on smartphone screen." Cheng Xin, Getty Images, 8 Oct. 2025, www.gettyimages.ca/detail/news-photo/in-this-photo-illustration-a-smartphone-screen-shows-the-news-photo/2239939746?adppopup=true.

Introduction

Following the release of Sora 2, OpenAI’s latest text-to-video AI model, public attention has once again returned to the ethical implications posed by generative AI. Sora 2, launched on September 30th alongside a respective iOS application, is capable of transforming user prompts into realistic short videos. Although the technology has been praised for its innovation and accuracy, it has also stirred widespread concerns regarding misinformation and copyright infringement.

The release of Sora 2 comes at a time when AI-generated video content is projected to account for 90% of online video by 2030, raising critical questions about authenticity and public trust (The Guardian). So how exactly will Sora — and AI-generated video content as a whole — impact the world in the near future?

How Sora Works

Sora 2 is a text-to-video generation tool, meaning it receives a user prompt as input and then generates a short video as output. Much like DALL-E, OpenAI’s text-to-image system, Sora-generated videos can mimic a multitude of styles, including realistic, animated, and hand-drawn. Text-to-video models are especially adept at understanding how the objects in a prompt behave in reality; thus, generated videos are capable of containing complex movements and emotions.

While OpenAI has declined to state which videos were used to train Sora, it is widely believed that the system has been trained on both copyrighted and publicly available media. Recent analyses estimate that major video-generation models train on over 800 million video clips, many of which include copyrighted material (Wall Street Journal).

What separates Sora 2 from previous versions — and from competing models — is its striking ability to generate hyperrealistic content. Most notably, it is OpenAI’s first text-to-video model that can generate videos with synchronized audio, including dialogue, sound effects, and diegetic noise. Sora 2 was also released alongside an iOS app of the same name, where users can create, upload, and share their own AI-generated videos. Despite being invite-only, Sora rapidly became the fastest-downloaded app on the App Store in 2025 (Axios).

Implications

Sora 2’s launch, paired with the AI-only Sora social platform, suggests that OpenAI intends to reshape generative AI’s role in social media. Rather than AI content being dismissed as “AI slop,” the company appears to be positioning AI-generated videos as a core content driver. This intention was made especially clear in Sora’s official launch video, where OpenAI CEO Sam Altman appeared in various AI-generated environments alongside Sora developers.

Some critics argue that text-to-video models serve limited practical purpose compared to chat-based models like ChatGPT. However, Sora will likely play a major role in media production; its low resource requirements make it an appealing and powerful tool for content creators. But given the hyperrealism and accuracy of Sora-generated videos, concerns have been raised regarding the ethical use of this technology.

Concerns

Watermark Removal & Metadata Vulnerabilities

Upon release, OpenAI implemented two measures to ensure Sora-generated videos remain identifiable:

• A large moving watermark

• AI-origin metadata attached to each file

However, metadata can be easily removed, and multiple third-party applications have already bypassed the watermark.

Copyright Infringement

Sora’s potential for copyright infringement has come under scrutiny, as protected characters and media have appeared in Sora-generated videos. Because Sora was trained on datasets presumed to include copyrighted material, some creators worry their work was used without consent. Although OpenAI updated its terms of service to allow rights-holders to license their content, critics argue this gives the company plausible deniability when violations occur.

In 2025, legal analysts estimated that over 40% of Sora-generated videos posted online contained copyrighted elements, highlighting the severity of the issue (Forbes).

Deepfakes and Disinformation

Given Sora’s realism, many fear that it will accelerate the spread of deepfakes. Videos of deceased cultural icons such as Michael Jackson and Robin Williams have already gone viral. Disinformation experts warn that deepfakes featuring living individuals — especially politicians — could manipulate public perceptions and sow confusion.

A report from AP News noted that deepfake incidents increased by 900% globally between 2022–2025, driven largely by consumer-grade AI tools (AP News).

As a result, individuals may face harassment, reputation damage, and fraud due to AI-generated impersonations.

Conclusion

For better or worse, AI-generated video content — including that produced through Sora — will become an inescapable form of media in the near future. It is therefore imperative that each major advancement in text-to-video models is matched by thorough regulation and rigorous safety measures by developers. At its best, Sora can empower anyone to create immersive and vibrant video content without extensive resources. At its worst, it risks eroding copyright protections and amplifying harmful disinformation.

Citations

“AI Video Will Make Up 90% of Online Content by 2030.” The Guardian, 2025, www.theguardian.com/us-news/2025/oct/04/openai-sora-violence-racism.

Axios Staff. “How to Spot a Sora Fake.” Axios, 12 Oct. 2025, https://www.axios.com/2025/10/12/spot-a-sora-fake.

Associated Press News. “Sora AI Video Generator Raises Disinformation Concerns.” AP News, 2025, https://apnews.com/article/sora-ai-video-generator-slop-openai-meta-ea4e4444bf90ca43c20a41b64b6716bf.

“OpenAI Reverses Stance on Copyright Works in Sora.” Wall Street Journal, 2025, https://www.wsj.com/tech/ai/openai-reverses-stance-on-use-of-copyright-works-in-sora-58d8c5fb

“Forbes Legal Entertainment: Sora 2 Does a Copyright Somersault Upon Launch.” Forbes, 17 Oct. 2025, www.forbes.com/sites/legalentertainment/2025/10/17/sora-2-does-a-copyright-somersault-upon-launch/.

New York Times. “Sora and the Future of Video Disinformation.” The New York Times, 2025, https://www.nytimes.com/2025/10/03/technology/sora-openai-video-disinformation.html.

OpenAI. “Sora 2 Overview.” OpenAI, 2025, https://openai.com/index/sora-2/.

OpenAI. “Generating Videos on Sora.” OpenAI Help Center, 2025, https://help.openai.com/en/articles/9957612-generating-videos-on-sora.

OpenAI. “Terms of Use.” OpenAI, 2025, https://openai.com/policies/row-terms-of-use/.