AI 기반 인지전의 현실: 이스라엘 경험이 시사하는 전략적 교훈
When Perception Becomes the Battlefield - Small Wars Journal
지정학적 인지전 통찰은 장기적인 전략적 맥락을 제공하지만, 즉각적인 방향성 움직임을 유발할 재정적 촉매제는 부족합니다.
핵심 요약
AI 기반 인지전은 공격 비용을 제로에 가깝게 만들었으며, 이는 미국 군사 전략에 중요한 대비책을 제시합니다.
핵심요약
- AI 기반 인지전의 비용은 거의 제로로 감소했습니다.
- 이란의 인지 작전은 독립적인 역량이 아닌 파비안 소모전 전략으로 기능했습니다.
- 이스라엘의 조직 적응 경험은 미국이 전쟁 긴급 상황 이전에 적용할 수 있는 교훈을 제공합니다.
도입
본 기사는 이스라엘이 AI를 활용한 인지전의 살아있는 실험장이 되었다는 점을 통해, 현재 서방 군사 전략의 인지전 개념이 실제 적용에 있어 얼마나 미흡한지를 보여줍니다. 투자자 관점에서 이 기사는 지정학적 위험이 기술적 역량과 어떻게 결합하여 비용 구조를 변화시키는지 이해하는 데 중요합니다. 즉, AI 기술의 발전이 군사적 개념의 정의와 실제 작전 간의 괴리를 어떻게 심화시키는지 분석합니다.
본문 1: 인지전 개념의 현실과 비용 구조
이스라엘의 경험은 AI가 어떻게 인지 공격의 비용을 근본적으로 변화시켰는지 명확히 보여줍니다. 기사에 따르면, AI에 의한 콘텐츠의 재맥락화는 인지 공격의 비용을 거의 제로에 가깝게 만들었습니다. 이는 공격 주체가 실제 정보 환경을 조작하고 재구성하는 데 필요한 인적 및 기술적 자원의 투입을 최소화했음을 의미합니다. 이는 전통적인 군사적 접근 방식이 정보전과 인지전의 복잡성을 포착하지 못했다는 점을 시사합니다. 즉, 공격 비용이 0에 수렴하면서, 정보의 진위 여부를 판단하고 대응하는 데 필요한 비용이 극적으로 감소했습니다. 이는 향후 군사 안보 예산이 정보 수집 및 분석보다는 AI 기반의 방어 및 대응 시스템 구축에 더 집중되어야 할 필요성을 제기합니다.
본문 2: 전략적 대응의 부재와 학습의 필요성
이러한 기술적 현실에도 불구하고, 미국 군사 내에서 인지전 개념에 대한 공통된 이해가 부족하다는 점이 중요합니다. 프랭크 호프만의 설문조사 결과에 따르면, 중국, 러시아, NATO의 연구에도 불구하고 미국 군대는 여전히 인지전에 대한 공통된 이해가 부족한 상태입니다. 이는 적대국들이 개념을 실제로 작전적으로 적용하는 동안 서방이 개념을 정의하는 데 시간을 소모했음을 의미합니다. 이스라엘의 사례는 이러한 인식의 격차가 실제 전쟁 상황에서 비용과 결과에 직접적인 영향을 미친다는 점을 입증합니다. 따라서 미국이 인지전의 이론적 정의를 넘어 실제 작전 환경에서 어떻게 적용하고 방어할지에 대한 실질적인 학습이 시급합니다. 이는 기술 선도국이 아닌, 실제 작전 환경을 경험한 주체가 전략을 주도해야 함을 의미합니다.
본문 3: 조직 적응과 미래 전망
이스라엘이 전쟁 상황에서 스펙트럼 작전, AI 분석, 민간 사회 모니터링을 통합하며 조직적으로 적응한 경험은 미래의 군사적 대응에 중요한 시사점을 제공합니다. 이러한 통합적 접근 방식은 분산된 정보 환경 속에서 생존하고 효과적으로 대응하기 위한 새로운 조직 모델을 제시합니다. 앞으로의 지정학적 경쟁에서 승리는 단순히 기술 우위를 넘어, 복잡한 정보 환경을 통합적으로 관리하고 민간 영역과의 경계를 허무는 조직적 유연성에 달려있습니다. 따라서 미국은 AI 기술을 활용하여 정보 통합 및 민간 사회와의 연계를 강화하는 조직적 변화를 가속화해야 할 것입니다. 이러한 적응력은 미래의 안보 환경에서 비용 효율성과 효과성을 동시에 달성하는 핵심 동력이 될 것으로 전망됩니다.
결론
결론적으로, AI 기반 인지전은 비용 구조를 재편하고 전략적 이해의 격차를 심화시켰습니다. 이스라엘의 사례는 기술적 진보가 실제 전쟁 환경에서 어떻게 작전적 현실로 변환되는지를 보여주는 살아있는 증거입니다. 투자자들은 이러한 인지전 역량이 향후 국방 예산 배분과 기술 투자에 미칠 영향을 면밀히 주시해야 하며, 조직적 유연성과 AI 통합 능력을 갖춘 국가의 전략적 우위가 장기적인 안보 환경을 결정할 것으로 예측됩니다.
Original Article
When Perception Becomes the Battlefield - Small Wars Journal
Israel’s experience as a live laboratory for AI-enabled cognitive warfare since October 2023 offers three findings for the Small Wars Journal (SWJ) community. First, AI-driven recontextualization of authentic content has collapsed the cost of cognitive attacks to near zero. Second, Iran’s cognitive operations served a Fabian attrition strategy rather than functioning as a standalone capability. Third, Israel’s forced organizational adaptation under fire, merging spectrum operations, AI analysis, and civil society monitoring, reveals lessons the US can apply before wartime urgency demands them.
Frank Hoffman’s November 2025 survey of the cognitive warfare concept in SWJ noted that despite a decade of Chinese, Russian, and NATO writing, “there is no common understanding of Cognitive Warfare” in the US military. The 2026 National Defense Authorization Act (NDAA) has since directed the Secretary of Defense to define it. Jeremiah Lumbaca’s Joint Special Operations University (JSOU) work, also discussed in SWJ , warned that the US is not yet prepared to face adversaries using cognitive warfare tactics. From where I sit in Israel, I can report that this assessment is correct, and that the cost of unpreparedness is visible in ways the doctrine has not yet captured.
This is not a theoretical argument. Since October 7, 2023, Israel has been a live laboratory for AI-enabled cognitive warfare. What Lumbaca called “cognitive contagions” and what Hoffman described as “weaponized social media with algorithmic amplification” have been deployed against Israeli society continuously , by adversaries who read the same PLA and NATO literature that Western militaries study. The difference is that adversaries applied the concepts operationally while the West was still defining terms.
What Israel’s experience adds to the cognitive warfare literature published in SWJ comes down to three observations. They are drawn from tracking the conflict’s information dimension from inside Israel, reviewing Israeli government and civil society monitoring data, and comparing observed adversary behavior against the frameworks that Hoffman, Lumbaca, Cheatham et al. , and others have published in SWJ.
The January 2026 piece in these pages on cognitive warfare economics focused on the defender’s cost asymmetry. Data from the Israel-Iran war extends that analysis. A study by the Israeli Internet Association , based on 592 fact-checks by 50 organizations across 23 countries, documented the scope of AI-generated disinformation during twelve days of fighting in June 2025. Roughly one-fifth of false content was AI-generated. But the more revealing figure is that 70 percent was authentic footage stripped of time, place, or framing context. The cost of producing a convincing cognitive attack dropped to near zero, not because fabrication got cheaper, but because recontextualization of existing footage reached industrial scale. As we have documented at FactSignal , this pattern mirrors what we tracked during the Gaza conflict’s first months, when recycled footage from Syria and Libya was repackaged as fresh Gaza coverage.
This matters for the cost asymmetry argument. Detection requires multilingual, cross-platform monitoring across Hebrew, Arabic, English, and Persian. Israel had only two active fact-checking organizations during the Iran war, FakeReporter and Bodkim, both publishing primarily in Hebrew. None appeared in Arabic. A country of nine million people was receiving cognitive attacks in four languages and could verify them in one. That gap is not about cost. It is structural.
Hoffman’s assessment noted that “the national security community is reducing its ability to monitor and respond effectively.” Israel’s experience suggests that the reduction is not merely budgetary. The monitoring infrastructure required to detect cognitive attacks at speed does not exist at sufficient scale in any single nation, including those facing active operations.
Lumbaca’s work published in SWJ in February 2026 described cognitive warfare as a strategy for “altering enemy cognitive processes” to achieve “political or military objectives.” Applied to Iran, this framing requires scrutiny. Did Iran pursue cognitive warfare as a primary strategy, or did cognitive operations serve a Fabian strategy of attrition and delay?
The evidence from the June 2025 war points to the latter. Iran’s forward defense was built on proxies meant to prevent confrontation on their own borders. When that architecture collapsed, with Hezbollah degraded and Hamas decapitated, the proxy ring failed. Iran was drawn into the direct war it had spent decades avoiding. Cognitive operations during the war were not a strategy of choice but a tool of necessity, deployed to compensate for conventional inferiority. This aligns with what the Belfer Center analyzed in its assessment of Iran ‘s proxy architecture unraveling.
The same Israeli Internet Association study found that 72 percent of false content served Iranian interests and 90 percent of AI-generated content did. Most Iranian disinformation focused on inflating perceptions of military power, what the study called “ballistic fakes.” Videos of missile trucks, fabricated strikes on Israeli infrastructure, and simulated destruction of Israeli landmarks. The content pattern reveals Iran is using psychological warfare techniques aimed at morale, not at decision-making processes or what Lumbaca terms “epistemic closure.”
Iran did not win a cognitive war. Iran lost a conventional one and used cognitive tools to manage the narrative consequences. This distinction matters for the SWJ community because it suggests that cognitive warfare, as Lumbaca and Hoffman define it, may be most dangerous not when adversaries choose it, but when they are forced into it as a substitute for capabilities they lack. The most aggressive cognitive actor may be the one losing on the physical battlefield.
This finding engages with the critique Matt Armstrong published in SWJ in November 2025, that cognitive warfare “fails the cognitive test” as a standalone concept. Israel’s experience suggests a partial answer. Cognitive warfare may not constitute a standalone strategy, but as a component of a broader asymmetric approach, accelerated by AI tools, it creates effects that current doctrinal frameworks are not built to handle at the speed they occur.
The US is not idle on cognitive warfare. The 2026 NDAA mandate to define the concept, the Special Operations Command’s (SOCOM) ongoing Psychological Operation (PSYOP) modernization , the Hudson Institute’s information inoculation training modules , and the Pentagon’s Operations in the Information Environment strategy all represent serious investment. A fair question is what Israel is actually doing that the US and its allies are not. The answer is less about technology and more about organizational adaptation under fire.
The IDF has restructured its C4I and Cyber Defense Directorate to create dedicated AI and spectrum warfare divisions . The new Sphera unit consolidates spectrum operations, strategic communications, and engineering teams under a single command. Rather than a PSYOP shop attached to a warfighting headquarters, the Sphera unit represents an organizational recognition that the electromagnetic spectrum, AI-enabled analysis, and narrative operations constitute a single battlespace. The US still treats these as separate functional areas under Joint Publication 3-13, Information Operations .
Israel has also developed a distributed monitoring network that integrates military and civil society actors. FakeReporter, an independent NGO, operates as the primary open-source intelligence and disinformation monitoring body in Israel. During the Gaza war’s first week, FakeReporter documented over 200 distinct disinformation narratives . The IDF refers to civil society monitoring in its situational awareness, and civil society monitors feed military intelligence. As opposed to formalized doctrine, Israel adopted the strategy out of necessity. It works because Israel’s small size and high digital penetration make the integration organic. The US, with a vastly larger information environment, would need a different architecture to achieve the same effect. We have tracked this civil-military integration at Allyvia, and it remains poorly understood outside Israel.
The battlefield has forced the IDF to treat AI-generated content as an immediate operational concern rather than a strategic abstraction. During the Iran war, AI-generated videos of Israeli F-35s being shot down , fabricated strikes on Ben-Gurion Airport , and synthetic footage of Israeli soldiers circulated within hours of actual events. The IDF’s response was not to build better detection tools, which remain six to twelve months behind generation capabilities. Instead, the operational adaptation was speed of authoritative response. The military learned to publish verified information, including corrections to its own claims, faster than adversarial narratives could solidify. This is closer to what Cheatham et al. described in these pages as “cognitive resilience as a critical weapon system,” though Israel arrived at it through battlefield iteration rather than institutional planning.
Israel’s experience does not yield a template. The US and its allies operate at a scale that makes Israel’s integrated civil-military model difficult to replicate. But three lessons transfer.
The monitoring coverage gap is the critical vulnerability. The US needs multilingual, cross-platform detection infrastructure that operates at the speed of adversarial content production. Israel’s two-organization fact-checking capacity was overwhelmed during twelve days of fighting. A country of 330 million people facing coordinated cognitive operations would be similarly exposed.
Organizational integration matters more than technology. Israel’s decision to merge spectrum, AI, and strategic communications under a single command reflects a lesson that this journal has been advancing for years. Hoffman noted the American “predisposition to kinetic operations” as a cultural barrier. Israel overcame that barrier because wartime urgency demanded it. The US should not wait for the same urgency.
And most relevant to Lumbaca’s call for cognitive warfare cells at all echelons, Israel’s experience confirms that distributed adaptation outperforms centralized response. The most effective Israeli responses to cognitive attacks during the war came from combinations of military intelligence, civil society monitors like FakeReporter and Bodkim, academic researchers, and independent analysts. Lumbaca recommended multi-disciplinary cells including data scientists and behavioral psychologists. Israel’s wartime version included journalists, NGO researchers, and open-source investigators. The principle is the same. Cognitive warfare cannot be defeated by a single agency. It requires what Lumbaca called “cognitive firewalls,” but those firewalls need to be societal, not merely military.
The cognitive warfare concept, as Hoffman observed, faces an uphill fight for acceptance within the US military. Israel’s experience since October 2023 suggests the fight is already over for anyone paying attention. The tools are deployed, the effects are measurable, and the cost of inaction is visible.
What is less clear is whether cognitive warfare constitutes a standalone domain, as the NDAA mandate implies, or whether it is better understood as a dimension of every existing domain. Israel’s data suggests the latter. Iran’s cognitive operations were subordinate to a Fabian attrition strategy, not independent of it. The most effective cognitive defenses were integrated into conventional military operations, not separated from them. The monitoring that mattered came from civil society, not from a dedicated cognitive command.
For the SWJ community, the Israeli case offers something the existing literature, strong as it is, has lacked. A sustained, real-world test of these concepts against an adaptive adversary using AI tools at operational speed. The results are still being analyzed. But the preliminary finding is clear enough. Cognitive warfare is not coming. It is here. The question is whether the institutional architecture to address it can be built in peacetime, or whether, as in Israel, it will require a war to force the adaptation.
Mark Ginsberg writes about AI, disinformation, and information warfare from Israel. His work has appeared in the Times of Israel, and he tracks cognitive operations at FactSignal and PaxPoint.
Unlock exclusive members-only ad-free content, members discussion, content, and updates directly from the SWJ Team, for only $10/yr .