Iran's AI Push Under Khamenei: What's Actually Happening
Iran isn't waiting for the West to hand it technology. Under Ayatollah Khamenei's direct guidance, the country has made artificial intelligence a national strategic priority, weaving it into everything from military doctrine to domestic surveillance infrastructure. This isn't speculation. It's a documented policy direction that's been building since at least 2019 and has intensified dramatically through 2025 and into 2026.
Khamenei has publicly framed AI as an instrument of national sovereignty. His statements have consistently positioned technological self-sufficiency as inseparable from political independence, particularly in the context of Western sanctions. The message to Iranian technocrats and military planners has been clear: develop it domestically, or find ways to acquire it from willing partners.
For geopolitical analysts, investors, and policymakers, this development deserves serious attention. We've spent considerable time tracking Iran's AI trajectory and consulting available open-source intelligence. Here's our honest assessment of where things stand.
The Strategic Framework Khamenei Has Built
Iran's AI development doesn't look like Silicon Valley's model. It's state-directed, militarily oriented, and deeply tied to the Islamic Revolutionary Guard Corps (IRGC). The Supreme Council of Cyberspace, which operates under Khamenei's authority, has been the primary body coordinating national AI policy.
In 2021, Iran released its National AI Strategic Document. By 2025, a revised version had expanded the scope considerably, with specific mandates for defense applications, economic modeling, and information operations. The document explicitly cites AI as a tool for "resistance economy" development, a phrase Khamenei's government uses to describe economic strategies designed to function under prolonged sanctions pressure.
Three areas have received the most investment:
- Surveillance and social control: Facial recognition deployment, internet traffic analysis, and social media monitoring tools aimed at domestic dissent.
- Cyber operations: AI-assisted hacking tools, disinformation automation, and infrastructure attack capabilities attributed to IRGC-linked groups.
- Drone and autonomous weapons systems: Iran's drone program, already battle-tested in Ukraine (via Russian procurement) and regional conflicts, is increasingly incorporating AI-guided targeting.
Where Iran Is Getting Its AI Capabilities
Sanctions complicate direct technology transfer, but they haven't stopped it. Iran has pursued several channels simultaneously.
China remains the most significant partner. Bilateral technology agreements signed between Tehran and Beijing provide cover for hardware transfers, including the kind of GPU infrastructure needed for serious AI training. Chinese firms have supplied networking equipment and computing hardware that ends up in Iranian defense-adjacent institutions, often through intermediary jurisdictions.
Domestic university research is genuinely impressive. Sharif University of Technology and the University of Tehran have produced AI researchers who compete credibly at international conferences. Iran's brain drain problem is real, but enough talent remains, particularly those who've returned or stayed for ideological or family reasons.
Open-source tools fill critical gaps. This is underappreciated. Because AI research is largely published openly, Iranian researchers can access state-of-the-art methodologies, fine-tune open models, and build functional systems without needing proprietary Western software. The international AI community's commitment to open science has, unintentionally, made sanctions on AI substantially harder to enforce.
The Surveillance Application: Domestic Control First
Khamenei's government has prioritized AI-powered surveillance above most other applications. The 2022 Mahsa Amini protests demonstrated both the threat that organized dissent poses to the regime and the government's determination to suppress it using technology.
Since then, Iran has expanded facial recognition cameras in Tehran and other major cities. Social media monitoring programs now use natural language processing to flag content in Farsi, Kurdish, and other languages spoken within Iran. Analysts tracking this through open-source intelligence have documented the procurement patterns.
For context on how AI tools are used to monitor and analyze geopolitical events like this, our article on the best AI geopolitical risk analysis tools in 2026 covers the platforms that analysts actually use to track state surveillance programs and regional instability.
Iran's AI in Military and Cyber Operations
This is where the global implications get serious. Iranian cyber units, particularly APT33 and APT34 (both linked to the IRGC), have been integrating AI tools into their operations for years. What's changed is the sophistication.
Phishing campaigns are now personalized at scale using AI-generated content. Social engineering attacks have become harder to detect because the language models producing them have improved. Iran has also been developing AI-assisted vulnerability discovery tools, which allow smaller teams to identify exploitable weaknesses in critical infrastructure faster than was possible previously.
On the kinetic side, Iran's Shahed-series drones, widely used in the Russia-Ukraine conflict, represent a proven export product. The next generation of these systems incorporates more autonomous navigation and targeting features. That's an AI application with direct battlefield consequences.
What the Geopolitical Risk Looks Like in Practice
For analysts trying to assess regional risk, Iran's AI development creates several specific complications.
First, attribution becomes harder. When AI tools are used to conduct influence operations or cyberattacks, tracing responsibility back to state actors requires more sophisticated counter-analysis. The volume of automated content Iran can produce has increased its ability to run plausible deniability campaigns.
Second, the drone-AI combination changes the calculus for regional adversaries. Israel, Saudi Arabia, and U.S. forces in the region are all adjusting their air defense strategies in response to the possibility of larger, more autonomous Iranian drone swarms.
Third, AI-powered economic modeling gives Iran better tools for sanctions evasion. By identifying gaps in international financial monitoring, optimizing cryptocurrency transactions, and modeling the behavior of sanctioning authorities, Iran can respond more quickly to economic pressure.
If you're an analyst or investor using AI tools to track these dynamics, our roundup of the best AI tools for geopolitical intelligence in 2026 covers platforms specifically built for this kind of structured risk analysis.
The Information Operations Dimension
Iran has been running sophisticated influence operations since at least 2016. By 2026, those operations are substantially more automated. Iranian state media and affiliated networks use AI content generation to produce articles, social media posts, and video content at a volume that would be impossible with human writers alone.
Tools like Jasper AI and platforms like Copy.ai represent the commercial end of AI content generation. State actors like Iran's information operations units have access to custom-built or open-source equivalents that are specifically tuned for their targets. The results show up as coordinated inauthentic behavior on social media platforms, fake news websites, and amplification of specific narratives in Western political discussions.
These aren't crude operations anymore. The targeting is precise, the content is linguistically sophisticated, and the distribution networks are designed to avoid platform detection.
Iran's Constraints Are Real Too
It would be wrong to overstate Iran's AI capabilities. There are genuine structural limitations.
Compute access remains a bottleneck. Training large AI models requires significant GPU infrastructure. Sanctions have made acquiring Nvidia's high-end chips genuinely difficult. China has helped, but the supply is constrained. Iran is working with older or less powerful hardware compared to leading AI nations.
Brain drain is a serious problem. Iran produces talented engineers and data scientists. A significant portion of them leave. The Islamic Republic's political environment, economic stagnation, and internet restrictions make it difficult to retain the people most likely to build cutting-edge systems.
Internet restrictions limit training data. Iran's national internet filtering, the "National Information Network," restricts access to vast swaths of the global web. This limits the data Iranian researchers can use for training models on open-domain tasks, even if it doesn't prevent them from using domestically generated data for specific applications.
How Analysts Are Tracking This
Open-source intelligence has become the primary method for tracking Iran's AI development given the closed nature of the programs. Researchers at institutions like the Atlantic Council, Bellingcat, and various university programs use a combination of academic publication tracking, procurement record analysis, and social media monitoring.
AI-assisted research tools have made this work faster. Tools like Perplexity AI help analysts quickly surface relevant research and cross-reference claims across sources. For structured intelligence gathering, platforms reviewed in our article on the best AI research assistants in 2026 offer capabilities that are directly applicable to this kind of geopolitical tracking work.
The challenge is separating credible signals from noise. Iran's government is also capable of deliberately seeding misleading information about its own capabilities, both to deter adversaries and to avoid revealing actual vulnerabilities.
What This Means for Western Policy
The existing sanctions framework was designed for a different era of technology transfer. Physical hardware is still constrainable to some degree, but software, algorithms, and research methodology are not. Western policymakers are grappling with the fact that export controls on AI are substantially harder to enforce than controls on centrifuge components or missile guidance systems.
Several specific policy debates have sharpened in 2025 and 2026:
- Whether to restrict access to open-source AI models for entities in sanctioned countries, and how that's even enforceable.
- How to work with semiconductor manufacturers and cloud providers to prevent compute access from reaching Iranian defense-linked institutions.
- Whether academic exchanges with Iranian researchers in AI fields should continue, given the dual-use nature of the work.
None of these have clean answers. The open science norms of the AI research community are in direct tension with national security objectives, and neither side is obviously wrong.
The Broader Pattern: Authoritarian AI Development
Iran's trajectory isn't unique. It fits a broader pattern of authoritarian governments treating AI as an instrument of state power, prioritizing control and military application over the kind of open, economically distributed development seen in democratic contexts.
Russia, China, and to varying degrees several other governments have made similar strategic choices. What makes Iran notable is that it's doing this under significant resource constraints and international isolation, which makes its progress more technically impressive even if the scale remains limited relative to the major powers.
For investors and analysts trying to model geopolitical risk, understanding which states are building AI capability independently, and how quickly, is increasingly central to assessing regional stability. The tools that help with that assessment are improving. But so are the capabilities being assessed.
Iran's AI development under Khamenei isn't a distant technical curiosity. It's a live geopolitical variable that's reshaping risk calculations across the Middle East and affecting the global information environment right now.
Our Assessment
Iran's AI program is real, growing, and consequential. It's not at the frontier of global AI development, but it doesn't need to be. For the specific applications Khamenei's government prioritizes, including domestic control, information operations, cyber capabilities, and autonomous weapons, the technology available to Iran is sufficient to cause significant harm and complicate adversary planning.
The constraints are also real. Compute limitations, brain drain, and data access restrictions all slow progress. But none of these are insurmountable, particularly with continued Chinese cooperation and the growing accessibility of open-source AI tools.
Anyone using AI for geopolitical intelligence work should have Iran's AI development on their radar as a variable that affects multiple domains simultaneously: cybersecurity, regional military balance, information integrity, and sanctions effectiveness.
For a broader view of how AI is reshaping geopolitical analysis and the tools analysts use to track developments like this, see our comprehensive guide to AI geopolitical risk analysis tools and our coverage of AI tools for geopolitical intelligence.