AI-Generated War Footage Spread on Social Media During Iran Conflict
AI-generated images purporting to show captured US soldiers and military defeats spread widely on social media platforms during the opening phase of the war in Iran in March 2026, featuring watermarks from Google's Gemini tool. The fake imagery was shared by prominent figures and circulated in multiple languages, fueled by limited journalist access and platforms' algorithmic prioritization of emotionally polarizing content.
What happened
AI-generated images purporting to show American soldiers captured by Iranian forces, B-2 bombers downed, and US military operations spread widely across social media platforms including X, Facebook, and Telegram during the opening phase of the war in Iran in early March 2026.
The images bore visual inconsistencies and watermarks from Google's Gemini AI tool, with captions claiming military victories for Iran. X had introduced rules to prevent its own AI system, Grok, from creating realistic images of current warzones, but users were still able to generate similar content using other methods.
Reported impact
Fake images circulated thousands of times and were shared by prominent figures, including Dr Jill Stein, a former US presidential candidate with 380,000 followers, who shared an image falsely claiming 173 American soldiers had been captured.
The spread was fueled by lack of journalist access to the conflict zone, polarized public opinion, and the emotional and polarizing content prioritized by social media platforms' engagement algorithms.
Index analysis
NewsGuard analysts identified X and Telegram as platforms where the most AI-generated conflict images were being shared. Many of the fake images observed pushed a pro-Iran perspective, claiming war success.
Content quality varied widely, from easily detectable low-quality "AI slop" to more convincing fakes. The ease of generating shock-factor imagery with AI contributed to rapid spread across multiple languages and regions.
Governance controls
X implemented rules restricting its Grok system from creating realistic images of current warzones, though the effectiveness of these controls was limited as users found alternative methods to generate similar content.
Platforms including X, Facebook, and Truth Social faced challenges addressing the surge of fake war imagery, particularly given their algorithmic prioritization of engaging and polarizing content.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.