Iran has mobilized tens of thousands of fake accounts to distribute deepfakes depicting its victory over Israel and the United States. In less than two weeks, this campaign orchestrated by the Iranian state reached 145 million views on social media, demonstrating that the cognitive warfare of militarized deepfakes now threatens global digital truth infrastructure on an unprecedented scale.
Cognition itself is now part of the battlefield. Deepfakes are being used as weapons to shape perceptions, obscure facts, and produce epistemic ambiguity. This instrumentalization reveals the progressive collapse of our collective verification systems in the face of militarized generative AI.
Iran Deploys the Largest State Campaign at Industrial Scale
The Iranian campaign deployed tens of thousands of synchronized fake accounts to distribute AI-generated deepfakes portraying Iran as victorious, its adversaries weakened, and its cause as legitimate. The activity centered around three recurring narratives designed to present Iran as the dominant and victorious actor in the conflict, amplified by synchronized publishing behavior and repeated media assets distributed across vast networks of accounts. The technological capacity to produce photorealistic and cinematically convincing synthetic videos has shifted from Hollywood studios to mobile devices operated by influence operations personnel.
Shortly after the launch of the joint American-Israeli military operation named “Epic Fury” on February 28, 2026, against Iranian nuclear facilities, military infrastructure, and leadership targets, a new wave of disinformation emerged, this time using AI-generated videos and images. The visuals were polished, notably, and in many cases entirely fabricated. AI-generated scenes of massive explosions in Tel Aviv, successful missile attacks on American warships, Israelis lamenting their losses, and other images purportedly showing how Iran inflicts damage on its enemies rapidly flooded social media. Many of the videos had a Hollywood quality, complete with massive explosions and sonic booms.
Documented examples reveal the sophistication of the operation. The New York Times identified over 110 unique deepfakes in the past two weeks conveying a pro-Iranian message through battlefield imagery, representations of missile strikes, and general war footage. As during the deluge of deepfakes in Iran’s June 2025 war, when Iranian accounts circulated false videos of Israeli landmarks in flames and recycled battle sequences from other conflicts, the objective of this content is to push a false narrative of Iranian military success and Western failure.
This campaign represents a critical technological shift. In 2023, an estimated 500,000 deepfakes were shared online. By 2025, this figure had reached 8,000,000, a 1,500% increase in two years. The 2026 Iran-United States-Israel war represents the moment when these technological and geopolitical trajectories converged with maximum force.
When Truth Infrastructure Collapses
The crisis extends far beyond deepfakes themselves. In a rapidly evolving conflict, verified information is often delayed, creating a void that disinformation immediately fills. When people are anxious, they seek information, but that information is often false. Unverified content can reach millions of people in minutes, and the public faces the difficult task of verifying content that is often highly realistic or shared across multiple sites.
Trust in visual media, the foundation of journalism, legal evidence, and social communication, is eroding. A 2025 Reuters survey revealed that 67% of respondents “often doubt the authenticity of video content,” compared to 33% in 2023. This “liar’s dividend”—where the existence of deepfakes allows real content to be dismissed as fake—could ultimately be more damaging than deepfakes themselves. These terms refer to a media landscape where AI-generated fake news casts even legitimate evidence into doubt, eroding trust to the point where any image or footage can now be rejected as a deepfake.
Current technological infrastructure reveals its limitations. The technology itself has improved notably, and it is now complemented by AI-powered chatbots integrated into search results and platforms like X. These chatbots have become new layers in the information ecosystem. Despite improvements, they still struggle to keep pace with real-time developments, but they are increasingly consulted as a first source of information. In the current conflict, Grok, X’s chatbot, flagged videos of Israeli Prime Minister Benjamin Netanyahu as deepfakes, which generated its own wave of confusion and fueled rumors about his whereabouts.
Deepfake Detection Falls Behind Generation
The deepfake detection landscape in 2026 resembles a high-stakes arms race. AI-generated images, videos, and audio have reached a quality level where human detection is essentially impossible. Early iterations were crude and easily identifiable, but modern deepfakes have achieved a level of photorealism and voice authenticity that can deceive even experienced observers and automated detection systems.
Traditional security systems struggle to keep pace with rapid improvements in deepfake models. Modern AI-generated videos can bypass detection tools with over 90% accuracy. Europe reveals its pocket giant strategy in the face of mega-raising in American AI, but this fragmented European approach struggles against the industrialization of deepfakes.
The industrial response remains insufficient. The technology industry rallied around the C2PA standard (Coalition for Content Provenance and Authenticity). Major camera manufacturers (Canon, Nikon, Sony) now integrate cryptographic provenance data at the moment of capture. Social media platforms (Meta, YouTube, X, TikTok) have implemented C2PA verification, displaying provenance information alongside content. Adobe’s Content Credentials system, integrated into Photoshop and Lightroom, creates an immutable editing history that tracks content across platforms.
But these solutions are reaching their limits. The C2PA system only works for content created with C2PA-compatible tools. Invisible watermarking uses patterns embedded in generated content that survive modifications but can be removed by sufficiently motivated adversaries with access to the detection model.
The Collapse of Institutional Safeguards
The scale of the crisis exceeds the capacity for self-regulation by platforms. X announced a policy suspending creators from its revenue-sharing program for 90 days if they post AI-generated conflict videos without disclosure, but this policy relies on metadata, which can be stripped, and on Community Notes attached as helpful to posts, which remain rare. X is just one of several platforms where real-time claims about the war propagate.
Researchers are not impressed. Joe Bodnar of the Institute for Strategic Dialogue told AFP that “the feeds I monitor are still flooded with AI-generated content about the war.” This saturation reveals the powerlessness of corrective measures in the face of state militarization of deepfakes.
Even if X’s demonetization policy were strictly enforced, a large number of X users distributing AI-generated content are not part of the revenue-sharing program. These users remain subject to verification through Community Notes, but the effectiveness of this community moderation tool is regularly questioned. Last year, a study by the Digital Democracy Institute of the Americas showed that over 90% of X’s Community Notes are never published.
With changes in platform partnerships, evolving funding models, and new technological challenges, many organizations need transition support to maintain their fact-checking work while building long-term sustainability. To support fact-checking newsrooms and promote information integrity worldwide, the Poynter Institute’s International Fact-Checking Network (IFCN) administers the Global Fact Check Fund, but these efforts remain marginal in the face of the industrialization of disinformation.
Toward Rebuilding Systems of Collective Truth
Even more advanced AI-generated disinformation is likely to be deployed in future conflicts, which could seriously compromise escalation management and public trust. Even then, however, deepfakes are likely to spread widely and shape broader perceptions of war. This evolution demands a fundamental rethinking of our approach to verification.
Deepfakes deceive and confuse the public and possibly even government officials, poisoning democratic discourse. Democratic governments should increase staffing dedicated to this issue and push technology companies to do the same. Information sharing is vital. Technology companies can learn how their platforms are being manipulated in ways governments must know about, while intelligence and security agencies can discover plans for deepfakes and other manipulations and warn technology companies.
China is reshaping the geopolitical balance of AI by exporting open source as new technological diplomacy, but this geopolitical fragmentation further complicates building global verification standards.
The solution requires a multilayered approach. Defense must be multilayered: identity verification, AI-powered media analysis, human oversight, and regulatory compliance. No single tool is sufficient. All panelists agree that while AI is a powerful tool, its effectiveness and reliability depend on careful management, transparency, and quality of underlying data. Journalists must be cautious about AI’s limitations and ensure they verify and fact-check information it produces.
The 2026 Iranian conflict: it was the largest conflict where AI-generated disinformation operated at scale, and it will definitely not be the last. The barrier to entry is now virtually nonexistent: anyone with a smartphone and a prompt can create convincing fakes in minutes. This accessibility of deception is both terrifying and fascinating. Between making creation tools accessible and militarizing cognitive manipulation, our societies must choose to collectively rebuild the foundations of digital truth.