More than 300 AI-generated videos of the Iran conflict have racked up tens of millions of views on X since fighting began on February 28. That alone would be a serious misinformation problem. But the platform's own AI chatbot, Grok, made it worse by repeatedly insisting fake content was real.
When users asked Grok to verify suspicious footage, it didn't just fail to flag the fakes. It actively vouched for them. In one case, Grok told a user that an AI-generated missile video was authentic, then fabricated citations from Newsweek and Reuters to support its claim. Those reports didn't exist. The chatbot invented sources to make its wrong answer look credible.
The Scale of the Problem
BBC Verify senior journalist Shayan Sardarizadeh said this conflict "might have already broken the record for the highest number of AI-generated videos and images that have gone viral during a conflict." The fakes range from dramatic to absurd: a viral clip showing Dubai's Burj Khalifa engulfed in flames, a fabricated satellite image of a U.S. radar facility in Qatar (confirmed as Google AI-generated), and AI images claiming to show the body of Iranian supreme leader Ayatollah Khamenei being recovered. Texas Governor Greg Abbott even reposted footage from the video game War Thunder as real combat.
X's Response Falls Short
On March 4, X's head of product Nikita Bier announced that creators posting AI-generated videos of armed conflict without disclosure would be suspended from the Creator Revenue Sharing Program for 90 days, with repeat offenders facing permanent removal. That addresses the financial incentive for posting fakes, but does nothing about Grok's verification failures.
The core problem here isn't just that AI-generated war footage exists on social media. That was inevitable. It's that the platform built a verification tool that confidently lies. Users asking Grok "is this real?" are doing exactly what you'd want people to do when they encounter suspicious content. Grok rewarded that skepticism with fabricated citations and false confidence. A chatbot that says "I can't verify this" would be infinitely more useful than one that invents Reuters articles to confirm fakes.
Until X fixes Grok's tendency to hallucinate sources during breaking news events, the chatbot is functioning as a misinformation amplifier, not a fact-checker.