IDF Faces Backlash and Internal Probe Over AI Targeting in Gaza
The documentary "NZA" exposes the IDF's use of AI in Gaza, sparking severe internal backlash and a rare intervention by Chief of Staff Eyal Zamir to protect Unit 8200.
The documentary film "NZA" has brought back to the forefront one of the most controversial issues surrounding the fighting in Gaza: the extent to which the IDF utilizes artificial intelligence to locate and identify targets. Until now, the main claims on the matter have been published in reports by foreign media, including The New York Times, which detailed how Unit 8200 integrated AI capabilities into its developed systems for intelligence purposes.
Among other things, in a revealed excerpt from the film, one of the interviewees stated that although the number of casualties resulting from "collateral damage" in strikes was not always verified, "approval was given for 500 [casualties]" in one of them. In a conversation with ynet, a senior military official dismissed the claim, explaining that "the figure of 500 is absurd." He noted that there are various levels of NZA, ranging according to approvals from senior echelons, but these do not reach the hundreds, only low dozens.
Many attempts to locate potential targets, combining artificial intelligence and facial recognition, originated from cooperation between Unit 8200 personnel and reservists working in technology companies such as Google, Microsoft, and Meta. Unit 8200 established what later became known as the "Studio"—an innovative center where experts are matched with AI projects. According to The New York Times, these cutting-edge technologies sometimes led to identification errors and the deaths of civilians in the Gaza Strip.
At the end of April 2025, a comprehensive article was published by The New York Times outlining the military's use of artificial intelligence technology in the war, particularly by Unit 8200. As an example, the American newspaper cited the targeted killing of Ibrahim Biari, the commander of Hamas's Jabalia Central Battalion, who was one of the leaders and planners of the October 7 massacre. According to the report, in late October 2023, Israel intended to eliminate Biari and knew he was hiding in the Strip's tunnel network, but intelligence failed to locate him.
Three Israeli and American officials told the Times at the time that Israel decided to turn to a new military technology integrated with artificial intelligence. That technology had been developed a decade earlier but had not been used on the battlefield until then. The attempt to locate Biari provided an incentive to improve the tool, and engineers in Unit 8200 integrated artificial intelligence into it.
Even if the IDF does use AI tools, this does not indicate anything flawed in the decision-making process regarding considerations in strikes and whether innocent people will be harmed. Every strike and targeted killing is examined at various echelons and passes through many eyes. Regarding strikes, many considerations are taken into account: whether the target even meets the criteria for a targeted killing, through a comprehensive investigation of its entire environment, its daily routine, the places where it is located, including surrounding locations, the type of armament, the nature of the operation, and additional considerations.
As part of the attempt to locate Biari, according to The New York Times, Israel listened to his communications and examined the same AI-based audio tool, which provided the estimated location from which he made the calls. Using this information, Israel launched an airstrike on the location on October 31, 2023, eliminating Biari. A human rights organization claimed that more than 125 civilians were killed in that strike.
The Times noted at the time that it conducted interviews with nine senior defense officials in Israel and the US, indicating that the assassination of Biari using that audio tool is just one example of how Israel is using the war in Gaza to test and rapidly deploy artificial intelligence technologies for military use on an unprecedented scale.
Cooperation Between Unit 8200 and Tech Giants
According to the report, in the 18 months preceding it, Israel integrated artificial intelligence into facial recognition software to compare blurred images with real identities, used it to compile a list of potential targets for airstrikes, and created an Arabic AI model to generate a chatbot capable of scanning and analyzing text messages, social media posts, and other data in Arabic. Many of these attempts were born out of collaboration between Unit 8200 personnel and reservists working at tech giants like Google, Microsoft, and Meta.
One of the tools developed by the "Studio," according to the Times, is an Arabic artificial intelligence model known as a "Large Language Model." The American newspaper noted that developers initially struggled to create such a model due to a lack of Arabic databases to train the technology. When data was available, it was primarily in Modern Standard Arabic, which differs from spoken dialects. However, the IDF did not face such a problem, thanks to decades of deciphering text messages, telephone call transcripts, and social media posts in spoken Arabic.
That Large Language Model was consequently created in the first months of the war, alongside the development of an Arabic chatbot. Unit 8200 integrated the tool with multimedia databases allowing analysts to run complex searches using images and videos.
The Chatbot's Errors
Following the elimination of Hezbollah Secretary-General Hassan Nasrallah, the chatbot analyzed reactions in Arabic-speaking countries. At times, the chatbot failed to recognize modern slang terms and words translated from English into Arabic. This required IDF intelligence officers specializing in various dialects to improve the system. The chatbot also occasionally provided incorrect answers—such as images of pipes instead of weapons.
Still, the artificial intelligence tool significantly accelerated various research and analysis efforts. In the temporary crossings established between the northern and southern parts of the Strip at the time, the Times claimed, Israel equipped cameras with the ability to scan and send high-resolution images of Palestinians to the AI-based facial recognition program. However, the system sometimes had difficulty identifying people whose faces were covered. This led to arrests and interrogations of Palestinians mistakenly identified as suspects.
Another use of artificial intelligence was to filter data on Hamas operatives gathered by intelligence. Before the war, Israel built an AI-based system known as "Lavender," which can quickly classify data to strike low-ranking militants. It included a database of verified Hamas operatives and was designed to predict who else might be a member of the terrorist organization. Lavender's assessments were not perfect, but Israel used the system at the beginning of the war to assist in target selection.
Ethical Questions
The cutting-edge technologies raise "serious ethical questions," Hadas Lorber from the Holon Institute of Technology (HIT) told the American newspaper back then. She warned that checks and balances are needed for artificial intelligence, adding that "human beings must be the ones to make the final decisions."
A spokesperson for the IDF stated more than a year ago to the Times that it cannot comment on specific technologies due to their "classified nature," clarifying that Israel is "committed to the lawful and responsible use of technological tools."
According to the Times, Israeli intelligence used the same audio tool it relied on to locate Biari, alongside maps and images of the underground tunnel network across the Strip, to locate hostages as well. Furthermore, two Israeli senior officials told the American newspaper that over time the tool was improved—making it more accurate for locating individual people.
The Unit at the Center and the Backlash
Against the backdrop of the film, there is an understanding within the IDF that immense damage has been caused to the system, prompting Chief of Staff Lieutenant General Eyal Zamir to personally intervene in an event he views not merely as a public relations issue, but as one that could significantly harm the activities of the IDF and the soldiers currently stationed on various fronts if not addressed.
The steps chosen by the Chief of Staff are highly unusual, among them targeting those who chose to be interviewed in the film, and the military is currently examining whether they exposed classified intelligence information prohibited from publication.
Even these days, the IDF's counter-terrorism efforts continue, primarily in the Gaza Strip, against the backdrop of American attempts to promote Trump's plan and IDF withdrawal. In many cases, activities are halted due to restrictions on harming innocents. Furthermore, according to security officials, the strikes currently taking place in the Strip are scrutinized under many eyes, and some are halted due to fears of losing legitimacy in global public opinion.
Just at the beginning of last month, forces in the Strip received a new directive stipulating that from that moment on, only strikes to remove immediate threats would be carried out. According to the directive, targeted killings are executed only with the approval of the Chief of Staff and the political echelon.
If it turns out that the soldiers interviewed in the film are currently serving in the IDF or the reserves, the military will examine whether and how to take action against them. IDF officials explain that the film causes immense damage to IDF operations across various arenas, with primary attention in terms of public relations and investigation naturally focused on Unit 8200, which is at the center of the film.
The statements made in the film have significant implications, as the unit operates with various units and entities worldwide. Its alumni manage business relations that affect Israel's economy and security. Therefore, the IDF is stepping up to defend itself, as well as to defend the internationally renowned unit.
