In the early days of the internet, spotting “fake” content was relatively straightforward—often limited to poorly Photoshopped images that were easy to identify. However, the digital landscape has drastically changed. Today, we are inundated with AI-generated videos and deepfakes that range from fabricated celebrity appearances to false disaster broadcasts, making it increasingly difficult to discern what’s real. This challenge is set to intensify with the rise of new AI tools that blur the boundaries between reality and fiction even further.
One of the most notable players in this space is Sora, an AI video tool developed by OpenAI. While Sora itself has been creating waves, its latest iteration, Sora 2, has taken the internet by storm as a viral social media app. Sora 2 is an invite-only platform designed like TikTok, offering a nonstop feed of AI-generated videos—all of which are entirely fabricated. Users scroll through an endless stream of hyper-realistic, fictional scenes that look convincingly real, yet none of the content is genuine. The author of the original article aptly described it as a “deepfake fever dream,” underscoring both its surreal quality and the risks it poses.
Technically, Sora videos stand out from other AI-generated content due to their high resolution, synchronized audio, and creative storytelling. Compared to earlier AI tools like Midjourney’s V1 or Google’s Veo 3, Sora’s output is significantly more polished and immersive. One of its most popular features, known as “cameo,” allows users to insert other people’s likenesses into AI-generated scenes seamlessly. This capability makes it frighteningly easy to create videos that appear authentic but are entirely fabricated. Such tools are incredibly powerful but also raise serious concerns about misinformation and manipulation.
Experts are increasingly worried about the potential misuse of Sora and similar applications. Because these tools democratize the creation of deepfakes, virtually anyone can produce convincing yet fake videos. This amplifies the risk of misinformation campaigns, character assassinations, and erosion of public trust. Celebrities and public figures are especially vulnerable to being impersonated or misrepresented. Industry groups, such as SAG-AFTRA (the Screen Actors Guild‐American Federation of Television and Radio Artists), have urged OpenAI to implement stronger safeguards to prevent misuse and protect individuals’ likenesses.
Detecting AI-generated content remains an ongoing challenge for tech companies, social media platforms, and users alike. However, it is not an impossible task. There are several methods and clues that can help you identify whether a video was generated by AI tools like Sora.
One key indicator is watermarking. Every video created on the Sora iOS app includes a watermark when downloaded—a white Sora logo shaped like a cloud that bounces around the edges of the video. This watermark functions similarly to TikTok’s watermark, serving as a visual cue that the content is AI-generated. Many AI companies are adopting watermarking as a way to help users and platforms quickly flag synthetic media. For example, Google’s Gemini “nano banana” AI model automatically watermarks its images to denote their AI origins.
While watermarking is a useful tool, it is not infallible. Static watermarks can be cropped out easily, and even moving watermarks like Sora’s can be removed using specific apps designed for watermark removal. This means that watermarks alone cannot be fully relied upon to determine the authenticity of a video. OpenAI CEO Sam Altman has acknowledged this limitation, stating that society will need to adapt to a world where anyone can create convincing fake videos, implying a broader need for vigilance beyond just watermark detection.
For those who want to dig deeper, checking metadata offers another layer of verification. Metadata is information embedded automatically into digital files at the time of their creation. It can include details such as the device used to capture or create the video, timestamps, location data, and other technical information. Importantly, AI-generated content often includes metadata tags that indicate its synthetic origin.
OpenAI participates in the Coalition for Content Provenance and Authenticity (C2PA), which means that Sora videos come with C2PA metadata credentials. These credentials can be verified through the Content Authenticity Initiative’s verification tool, which allows users to upload a photo, video, or document and inspect its metadata for signs of AI generation. When a Sora video is checked using this tool, it typically shows that the content was “issued by OpenAI” and clearly indicates