VENICE/JERUSALEM, Sept. 12, 2026 — A new documentary built around testimony from 24 Israeli military and intelligence insiders has reopened one of the most consequential questions surrounding modern warfare: how far artificial intelligence and automated data systems have shaped Israel’s targeting campaign in Gaza — and where human responsibility begins and ends.
The Israeli military has categorically rejected key allegations in NAZA, an 80-minute documentary directed by Israeli filmmakers Yuval Abraham and Rachel Szor. The film premiered at the Venice International Film Festival on Sept. 10 and received a roughly 25-minute standing ovation.
A film built around anonymous Israeli insiders
NAZA was filmed largely at night on Tel Aviv rooftops over about three years. Its 24 military and intelligence sources appear with their identities concealed and their voices and appearances digitally altered.
According to the filmmakers, the insiders describe surveillance, target identification and bombing procedures they say helped turn Gaza into an increasingly data-driven battlefield. Several interviewees characterize their work as highly remote and industrialized, while the documentary argues that expected civilian casualties were sometimes incorporated into decisions before strikes occurred.
The title itself refers to an Israeli military shorthand for anticipated civilian or “collateral” casualties associated with a strike. The Venice Film Festival lists Abraham and Szor as directors and describes the film as an examination of Israeli military systems and civilian deaths in Gaza.
Abraham and Szor were also among the four filmmakers behind No Other Land, which won the Academy Award for best documentary feature in 2025.
The allegation at the center of the controversy: AI-assisted targeting
One of the most important elements of NAZA concerns Lavender, a military intelligence tool that has been the subject of reporting since 2024.
The documentary’s sources describe a system using large quantities of communications, movement and surveillance information to help identify people suspected of ties to Hamas or Palestinian Islamic Jihad.
Reuters reported that officers interviewed for NAZA described an AI-assisted process that processed information from large numbers of mobile phones and helped identify potential targets. The documentary argues that such systems greatly accelerated the ability to locate and strike suspected militants, including when they were in residential buildings.
But that is where an important distinction must be made.
Israel does not deny using advanced data-analysis tools in its intelligence work. It strongly disputes the claim that artificial intelligence independently selects people for attack or makes the final decision to strike.
IDF: Humans make the decisions, not artificial intelligence
Following the film’s release, the Israel Defense Forces said it rejected the allegations and questioned whether the anonymous interviewees could be independently verified as having served in the positions they claimed.
The military said some of the testimony appeared to involve strategic matters beyond the normal knowledge of junior personnel.
Most importantly, its official response states that decisions concerning the selection and approval of strikes are made by human personnel under military guidelines and “not by artificial intelligence.”
The IDF also says it attacks military targets in accordance with international humanitarian law, takes measures intended to reduce civilian casualties and investigates suspected violations when credible evidence emerges. It continues to accuse Hamas of locating fighters and military infrastructure inside densely populated civilian areas.
That denial means descriptions such as “AI-powered killing” should not be presented as an independently established fact. The more accurate characterization is that NAZA alleges extensive AI-assisted or algorithm-supported targeting, while the IDF says humans remain responsible for determining and authorizing attacks.
Lavender was controversial years before ‘NAZA’
The documentary is not the first time Lavender has come under scrutiny.
In April 2024, Israeli-Palestinian publication +972 Magazine reported, citing six Israeli intelligence sources, that Lavender was used to generate or identify tens of thousands of suspected militant targets during the early stages of the war. The investigation alleged that human review of some recommendations was extremely limited.
The Guardian subsequently reported similar accounts, while also publishing the IDF’s rebuttal. Israel said Lavender was not an autonomous target-selection machine but an information database designed to organize and cross-reference intelligence for human analysts.
In a more detailed June 2024 explanation, the IDF acknowledged the use of systems including Habsora, or “The Gospel,” and Lavender, but said Habsora recommendations require independent examination and that Lavender is a database rather than a list of people automatically approved for attack.
That distinction has remained central to the controversy ever since: critics argue that rapidly generated algorithmic recommendations can still shape deadly decisions even when a person technically gives final approval; the military argues that human review and legal procedures prevent machines from making those decisions themselves.
Human Rights Watch raised a different concern: automation bias
Human Rights Watch examined Israeli military digital tools in 2024 and warned that machine-learning systems could increase risks to civilians if their outputs were treated with excessive confidence.
The organization did not claim it could independently document every instance in which the tools were used. Instead, it argued that systems such as Lavender and Habsora could produce errors because algorithms depend on incomplete data, assumptions and the criteria chosen by their developers.
That issue — often described as automation bias — may ultimately be more important than the narrower question of whether an algorithm literally “pressed the button.”
A human can technically remain in the decision loop while relying heavily on machine-generated recommendations.
The civilian toll keeps the debate from being purely technological
The argument over artificial intelligence is unfolding against the enormous human cost of the Gaza war.
CNA, citing Gaza’s health ministry, reported at least 73,658 Palestinian deaths since October 2023 as of Sept. 11. The ministry does not provide a comprehensive distinction between militants and civilians in its overall total. CNA reported that the United Nations regards the ministry’s aggregate casualty figures as credible, while Israel has repeatedly challenged interpretations of those numbers and says Hamas’s operations from civilian areas contribute to the scale of casualties.
The war followed the Hamas-led attack on southern Israel on Oct. 7, 2023, in which roughly 1,200 people were killed and 251 people were taken hostage, according to Israeli figures cited by major international news organizations.
These competing narratives are critical context. NAZA presents testimony alleging that civilian deaths were not simply unavoidable consequences of urban warfare but were sometimes knowingly tolerated as part of a broader targeting system. Israel rejects that characterization and says its operations target militants while seeking to minimize civilian harm.
Why ‘NAZA’ could resonate far beyond Gaza
The controversy is bigger than one documentary and one military.
Armed forces worldwide are investing in artificial intelligence to process surveillance, identify patterns, prioritize threats and accelerate what military planners call the “kill chain” — the sequence from detecting a potential target to deciding whether to attack it.
The unresolved question is what meaningful human control actually means.
If an algorithm analyzes millions of data points, assigns risk scores and recommends targets while a human officer approves the result, is the officer genuinely making an independent decision — or merely validating a machine-generated conclusion?
That question is likely to become increasingly important as militaries integrate AI into intelligence and targeting systems.
For now, NAZA offers testimony from people who say they saw one version of that future from inside Israel’s military apparatus. Israel says their account is unverifiable, misleading and fundamentally wrong about how targeting decisions are made.
The documentary therefore does not settle the argument.
It makes the argument much harder to ignore.
Leave a Reply