AI-Generated Videos Spread Misinformation on Social Media Platforms
Since OpenAI's Sora video generation tool launched two months before December 2025, AI-generated videos designed to deceive have proliferated across major social media platforms including X, YouTube, Facebook and Instagram. Despite watermarking and content moderation policies, users have found ways to remove watermarks and spread misinformation, including in foreign influence campaigns, with platforms struggling to detect and label synthetic content effectively.
What happened
In the two months since OpenAI's Sora was released, deceptive AI-generated videos have surged on X, YouTube, Facebook and Instagram, according to experts who track them. While many videos are silly memes or fake images of babies and pets, others are intended to stoke vitriol in political debate and have been used in foreign influence operations.
People with malicious intent have discovered ways to circumvent disclosure rules by manipulating videos to remove watermarks, with several firms offering services to strip logos. Both Sora and Google's rival tool Veo embed visible watermarks and invisible metadata to signal AI origin, but these protections have proved inadequate.
Reported impact
Deceptive videos have raised alarm over a new generation of disinformation and fakes. The tools have become sophisticated enough to fool millions of people, blurring the line between reality and AI-generated content.
Index analysis
Researchers noted that social media companies' policies requiring disclosure of AI use and prohibiting deceptive content have proved woefully inadequate for the technological leaps these tools represent. Sam Gregory, executive director of Witness, stated that platforms could do better in content moderation for misinformation and in proactively labeling AI-generated information.
YouTube uses Sora's invisible watermark to append labels indicating AI videos were altered or synthetic, but more work is needed across platforms. OpenAI said it prohibits deceptive uses of Sora and takes action against violations, but enforcement appears insufficient given the scale of deceptive content.
Governance controls
Major social media companies have policies requiring disclosure of AI use and broadly prohibiting content intended to deceive. OpenAI and Google embed visible watermarks and invisible metadata in their video tools to signal AI origin to users and platforms.
YouTube has implemented technology to automatically detect and label AI-generated videos using Sora's invisible watermark.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.