Just a few years ago, neural networks could barely produce short clips, but by 2026 they had become part of information warfare. They are now used to create convincing images and videos that spread rapidly on social media without the need for massive bot farms or multimillion-dollar budgets.
Such material is increasingly described as slopaganda. AI ethics researchers Michal Klincewicz, Mark Alfano, and Amir Fard argue that it differs from older fake propaganda in its scale and its ability to target both broad and highly localized audiences through pseudo-news websites.
Research suggests that audiences struggle to identify these fakes. In a Stanford study from 2024, about 43.5% of respondents believed propaganda materials created with the then-current version of ChatGPT-3. In a University of Chicago experiment the same year, only 40% correctly identified AI-generated images. Younger users detect fakes more successfully, while older people have more difficulty. Polls in 2025 found that about 25% of Americans generally trust chatbot-generated information without checking it, and 9% of U.S. residents consume news through AI services.
“Hundreds of thousands of hours of video are created every day... and that muddies and confuses our information environment,” visual communication professor TJ Thomson told ABC News.
One of the clearest examples in the article is Iran. According to Cyabra, during two weeks of conflict with the United States and Israel, tens of thousands of bots spread AI videos claiming major Iranian military successes. Those clips, including scenes of strikes on sites in Gulf countries, gathered more than 145 million views. Mark Jones of Northwestern University in Qatar told The New York Times that such footage helps portray the war as more destructive and more costly for U.S. allies.
The problem is compounded by failures of digital platforms and AI systems themselves. Google AI Overviews treated generated images of destruction as real, while Grok cited nonexistent Reuters, CNN, and Euronews reports. The text says that only X had promised penalties for posting AI videos without disclosing their origin.
The article also focuses on Russia. At first, pro-Kremlin networks circulated staged clips and footage from video games, but with the rise of generative tools they shifted to two formats: crude satirical videos aimed at demoralization and more realistic clips that can be mistaken for genuine footage. The story says that Sora 2 was used for some of this content, including videos claiming that Ukrainian soldiers were surrendering, retreating, or crying after forced mobilization. According to NewsGuard, the Kremlin-linked “Pravda” network has published more than 3.6 million AI-generated news items since 2024.
The article concludes that slopaganda is likely to remain a permanent feature of modern conflict: generators are becoming cheaper and more persuasive, while technology companies are doing little to slow its spread. As a result, users are finding it increasingly difficult to distinguish reliable reporting from fabricated content or to trust the news at all.