AI-generated videos
are becoming ubiquitous, ranging from celebrity deepfakes to false news broadcasts and trending clips of animals. OpenAI’s Sora, an AI video creation tool, complicates the task of distinguishing reality from fabrication. The latest version, Sora 2, continues to enhance its capabilities.
In recent months, the app, reminiscent of TikTok, has gained immense popularity with AI aficionados scrambling for invitation codes. However, Sora is distinct as it solely hosts AI-crafted content. It has been described as an
AI deepfake spectacle
that appears harmless but poses serious risks.
Stay updated with unbiased tech news and lab-based evaluations. Add CNET as a preferred source on Google.
Technically, Sora’s videos stand out against competitors such as
Midjourney’s V1
and
Google’s Veo 3
, incorporating high resolution, synchronized sound, and notable creativity. One of Sora’s standout features, termed “cameo,” allows users to integrate others’ likenesses into nearly any AI-generated scenario, yielding disturbingly realistic videos.
Concerns about Sora are widespread among experts due to its ability to facilitate the creation of malicious deepfakes and misinformation, making it harder to discern reality. Public figures are particularly at risk, prompting organizations like
SAG-AFTRA
to advocate for more robust safeguards from OpenAI.
Detecting AI-generated content remains a challenge for tech firms and social media networks alike. However, with some diligence, users can look for signs indicating a video was produced with Sora.
Identify the Sora Watermark
Every downloaded Sora video features a watermark—a bouncing white cloud icon—similar to TikTok’s. This watermark helps identify AI-generated content. Other AI tools, like Google’s Gemini, also employ watermarking for transparency.
Review the Metadata
While examining a video’s metadata might seem daunting, it is a great tool for confirming if it was created using Sora. Metadata includes details like the method of creation, date, and even AI origin.
Utilize AI Labels and Add Your Own
Platforms like Instagram and Facebook have systems that flag and label AI-generated content. While relying solely on these labels is not infallible, creators disclosing their content as AI-generated can aid in clarifying its source.
Maintain Awareness
There is no one-size-fits-all way to identify AI videos instantly. It’s crucial to approach online content with a critical eye. Whether it’s checking for nonsensical text or odd movements, enhanced scrutiny can protect against being misled. Remember, even experts make mistakes sometimes.
