The 2026 Iran war reveals how modern conflict is shaped by fused intelligence—HUMINT, technical surveillance, and AI—accelerating warfare while exposing new vulnerabilities in escalation control and decision-making
The Tomahawk missile explosions ripping through Tehran’s Pasteur Street on the morning of 28 February 2026 heralded the dawn of a new geopolitical age. In the days and hours since that first impact, the war has expanded into a wider regional and likely global conflagration, upending the existing regional order and deepening uncertainty about its long-term implications.
Sustaining this war—and perhaps holding the key to its resolution—is intelligence. It is embedded in data points appearing on command-and-control screens across the region, gathered on the ground through human agents, and increasingly collected, processed, and even independently acted upon by large language models (LLMs) and autonomous systems. Through an exploration of these features, this analysis examines how intelligence is shaping the trajectory of the ongoing hostilities and what its use in this context might reveal about its present and future role in modern statecraft.
The effectiveness and lethality of the Mossad’s covert operations against Iran and its regional proxies have delivered deep strategic advantages for Israel in its shadow war against Tehran through their psychological effect. The paranoia this has injected within Iran’s counterintelligence services, particularly over the past year, has eroded conventional deterrence. This absence of guardrails encourages further—and unpredictable—escalation in the war.
While fear of infiltration by hostile intelligence services has long coloured the Islamic Republic’s political culture since its inception, these concerns have mutated into a corrosive paranoia over the past decade amid the growing sophistication and frequency of Israel’s targeted assassinations of high-value figures within the country itself. The assassination of Mohsen Fakhrizadeh in 2020 was one critical juncture, with initial reports by the Defence Ministry suggesting that the nuclear scientist was killed in an alleged shootout between his bodyguards and armed terrorists, suggesting deep Israeli penetration of Iran’s security services and contradicting the later account put forth by the state. During and after the ‘12-Day War’ in the summer of 2025, Israel assassinated high-value targets deep inside Iran with increasing ferocity and regularity. Ismail Haniyeh, Hamas’ political leader, who was targeted by a suitcase bomb planted inside his Tehran safehouse in July 2024, is perhaps the best-known example. Incidents such as these, alongside the Mossad’s proven ability to infiltrate supply chains in ‘Operation Grim Beeper’ in September 2024 against the Lebanese Hezbollah, and short-range drone attacks by Mossad operatives across Iran, have only deepened this sense of paranoia within the country’s counterintelligence organisations.
This absence of guardrails encourages further—and unpredictable—escalation in the war.
The atmosphere of mistrust within Iranian counterintelligence has three major implications in light of the recent hostilities. First, as Iranian intelligence officials turn on each other, there will be greater demands for demonstrations of “loyalty” within both the leadership and the rank and file. This may encourage violent and risky actions, including heightened grey-zone warfare, threatening to further aggravate the ongoing conflict.
Second, concerns about hacking by Israeli intelligence have precipitated a return to slower analogue mechanisms within Iran’s national security systems—mechanisms that, although more immune to interception, are outpaced by the accelerated pace of modern warfare and the need for faster decision-making in crises. This has created a disconnect between the interim government in Tehran and the rank-and-file of the Islamic Revolutionary Guard Corps (IRGC), military, and intelligence services. Given the slow and often absent communication between the two, the latter now possesses greater latitude to act independently and escalate conflict beyond established thresholds, as Iran’s Foreign Ministry has already claimed. This is reflected in recent Iranian missile strikes on Saudi oilfields, the UAE’s metropolises, and even Turkey, despite the adversarial relationship with Israel that it shares with the latter.
Third, the ongoing purges of Iran’s intelligence services and the fear of being perceived as ‘moderate’ reduce the scope for clandestine diplomacy through intelligence-led backchannels to resolve conflict. Clandestine diplomacy is a core responsibility of intelligence services during war. Yet if internal dynamics disincentivise the pursuit of dialogue under the threat of probable execution, individuals within Iran’s intelligence services will be less likely to explore avenues of negotiation with their counterparts in the Central Intelligence Agency (CIA) or the Mossad, limiting the prospects of ending hostilities in the foreseeable future.
The uncharted trajectory of the war at this stage owes much to the assassination of Iran’s Ayatollah Ali Khamenei at its very beginning by US and Israeli forces. Yet the planning leading up to the war is equally important, revealing the integrated nature of human intelligence (HUMINT) and technical/signals intelligence (TECHINT/SIGINT) within the modern intelligence landscape. Most importantly, however, it punctures myths about the supposed impracticality or demise of HUMINT in a more ‘transparent’ operational environment.
HUMINT and TECHINT are no longer distinct from one another, as they may have been when modern intelligence bureaucracies were first developed over a century ago, but are increasingly intertwined.
The stages of planning demonstrate an extraordinary degree of coordination between HUMINT, ubiquitous technical surveillance (UTS), and strategic communications by both the CIA and the Mossad. First, it is reported that the Mossad had access not only to the electronic devices owned by Ayatollah Khamenei’s inner circle for nearly two decades but also to nearby mobile communications towers, enabling the denial of cell-phone signals and delaying early warning to the Iranians until the last minute. Second, Israel’s hacking of Tehran’s extensive network of surveillance cameras allowed it to leverage UTS—otherwise seen as a hindrance to HUMINT—in its favour, while also enabling intelligence officials to monitor the Ayatollah’s ‘pattern of life’ in the lead-up to the strike. Third, the presence of human operatives provided the Mossad and CIA with the knowledge and intuition that ultimately ensured the success of the strike—namely, information that the Ayatollah felt least vulnerable during daylight hours, that he would be arranging a meeting at the presidential palace with his inner circle, and even photographic evidence of his death in the immediate aftermath for the US and Israeli leadership, if Israeli media reports are to be believed. Finally, the deft use of strategic communications in the form of a well-timed ‘leak’ on the night of 27 February to the Israeli press alleging that Chief of the Israeli Defence Forces (IDF), Gen. Eyal Zamir, was absent from his office limited challenges from an open-source intelligence (OSINT) perspective by ensuring that the Israeli Air Force (IAF) maintained the element of surprise and that news of the impending strike did not reach broadcast or social media first.
All of this reveals a central truth about intelligence in contemporary warfare and statecraft: HUMINT and TECHINT are no longer distinct from one another, as they may have been when modern intelligence bureaucracies were first developed over a century ago, but are increasingly intertwined. HUMINT-derived knowledge about the Ayatollah’s sense of vulnerability and schedule was used to corroborate TECHINT-derived inputs informing pattern-of-life analysis by the Mossad and the CIA, and vice versa. More centrally, it underscores the enduring importance of HUMINT in a changing, technologically driven intelligence landscape. Factors such as the timing of the strike, its location, and its proof—all of which structured the strategic advantage enjoyed by the US and Israel in its aftermath—were all dependent on OSINT. Evidence such as this demonstrates that reports of HUMINT’s death, as often noted in recent work on the discipline, remain greatly exaggerated. To borrow an analogy from economics, HUMINT will likely be treated as a Veblen good going forward: despite the rising ‘costs’ of its action, given the emergence of UTS and advanced OSINT tools, demand for this form of intelligence activity will only grow among policymakers because of its inherent value in discerning the motivations, fears, and vulnerabilities of adversaries.
The use of intelligence in this war demonstrates the ubiquity of agentic artificial intelligence (AI) in modern warfare, with autonomous weapons platforms and large language models (LLMs) collecting data, processing it, and even executing commands independently of human guidance. The apparent decline of human-in-the-loop feedback cycles and kill chains naturally raises concerns about the future of intelligence. Yet the war has also exposed vulnerabilities associated with overreliance on AI in intelligence synthesis and operations, offering a glimpse of future intelligence priorities in a technology-driven security landscape.
Yet overreliance on AI as a mechanism for autonomous intelligence fusion, without adequate human supervision, risks damage to military platforms as well as casualties.
Despite the ongoing dispute between AI firm Anthropic and the Pentagon over proprietary use of its models, Anthropic’s Claude code has been employed by the US Air Force (USAF) in this war to synthesise data and intelligence while identifying bombing targets in real time. This has significantly shortened decision cycles, accelerated navigation and the development of targeting patterns, and ultimately enabled US and Israeli forces to achieve objectives more quickly.
However, overreliance on AI also expands the scope for vulnerability, as the war has shown in its early stages. The enduring Clausewitzian feature of the ‘fog of war’ remains impenetrable, as demonstrated by a case of ‘friendly fire’ between allied Kuwaiti air defences and US F-15 fighter jets on 2 March. Coordination of air defences and airstrikes is controlled by the US Combined Aerospace Operations Centre (CAOC) at Al-Udeid Base in Qatar, with this ‘blue-on-blue’ incident explained by experts as resulting from a “software problem,” rather than “human error or operator” fault at CAOC or in Kuwait. It may therefore be surmised that the incident arose from overreliance on agentic AI coordination systems, which failed to disaggregate the various complicating aspects of aerial warfare during a serious crisis. More centrally, the USA’s recognised dependence on AI models such as Claude has led Iran to strike the Amazon Web Services (AWS) data centres in the UAE on which it depends, aiming to ‘blind’ US forces and thereby undercut its strategic advantage.
Developments such as these highlight two key aspects of the AI-driven use of intelligence in this war. First, US and Israeli mastery of this technology and its seamless integration into autonomous systems and intelligence platforms has undoubtedly provided them with a strategic edge against Iran by collapsing decision cycles, synthesising data and intelligence across multiple platforms, and intensifying time pressures on Tehran. Yet overreliance on AI as a mechanism for autonomous intelligence fusion, without adequate human supervision, risks damage to military platforms as well as casualties. AI-based intelligence fusion systems also depend on physical infrastructure within the region—which, once ‘knocked out,’ can blind military command and control systems amid the fog of war.
With the war still recent, its full impact — in the intelligence sphere and beyond — has yet to fully crystallise. Yet its transformative effects on the global intelligence landscape are clear. Policymakers across the world should pay attention as they brace themselves to survive and even thrive in the dangerous world we now inhabit.
Archishman Ray Goswami is a Non-Resident Junior Fellow with the Observer Research Foundation and a DPhil International Relations candidate at the University of Oxford.
The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.
Archishman Ray Goswami is a Non-Resident Junior Fellow with the Observer Research Foundation. His work focusses on the intersections between intelligence, multipolarity, and wider international politics, ...
Read More +
PREV
