Introduction

The rapid diffusion of generative artificial intelligence has turned human cognition into a contested strategic domain. International Relations (IR) scholarship increasingly conceptualizes cognitive warfare as the deliberate shaping of perception, belief and decision-making, and links it to the geopolitics of chips, cloud, data and platforms. Yet this literature remains centered on the United States, China, Russia and NATO, and offers few theory-informed accounts of how sanctioned, non-Western states defend their information spaces. Iran is a conspicuous gap: its internet throttling, localization policies and the National Information Network (NIN) have been studied mainly as instruments of censorship or economic policy, rarely as components of a national cognitive-security strategy. This article asks how, and with what consequences, the NIN has functioned as cognitive-security infrastructure under sanctions and during the crisis episodes of 2019-2025. The hypothesis is that, through a securitizing move that equates the information space with sovereign territory, the NIN has been progressively transformed from a connectivity project into a layered instrument of cognitive sovereignty whose security gains are systematically accompanied by legitimacy and economic costs.

Methods

The study adopts a qualitative, theory-guided design with three components. First, a structured review of a 94-record literature corpus on AI-enabled cognitive warfare, AI geopolitics, digital sovereignty and Iranian security establishes the research gap. Second, process tracing is applied to three crisis episodes: the November 2019 unrest, the 2022 nationwide protests and the June 2025 Iran-Israel war. Each episode is coded on an ordinal scale (0-4) across four control instruments. Third, a structured, focused comparison situates Iran relative to the European, Chinese and American models of digital sovereignty along five dimensions. The analytical framework, termed cognitive-security sovereignty, integrates securitization theory, the bordering approach to digital sovereignty and Iranian strategic culture.

Results and discussion

The review shows that only 16 of 94 records (17%) focus on Iran and none directly examines AI-enabled cognitive operations in Persian-language crisis environments. Process tracing reveals a clear escalation and sophistication of control: the composite control index rises from 2.75 in 2019 to 3.25 in 2022 and 4.00 in 2025. More importantly, the logic of control shifts from blunt disconnection toward staged throttling, protocol suppression and whitelisting combined with domestic platform substitution, so that the NIN increasingly operates as a cognitive border rather than an on/off switch. The comparative analysis places Iran closest to the Chinese security-first model on content control and localization, but with markedly lower infrastructural autonomy and external interoperability, a combination shaped by sanctions. The result is a sovereignty-legitimacy dilemma: the instruments that buffer society from hostile narratives and preserve the continuity of domestic services also generate household economic losses, widespread circumvention and erosion of epistemic trust, which adversaries can exploit.

Conclusions

The NIN should be understood neither as a purely technical project nor as mere censorship, but as a sanctions-conditioned form of cognitive-security infrastructure whose effectiveness depends less on the depth of control than on the legitimacy of its use. The article contributes a non-Western case and a transferable framework to IR debates on cognitive warfare, and recommends a shift from disconnection-centered defense toward resilience built on transparency, legal time limits, verified domestic information hubs and AI-assisted detection of synthetic content.