Pursuing AI leaks in Gaza reveals the fragility of human oversight over targeting algorithms

The warm applause that lasted twenty-five minutes at the Venice International Film Festival turned into a security and political crisis inside Israel after Prime Minister Benjamin Netanyahu called for a comprehensive criminal investigation into the sources of the documentary “Naza.” The film, which won the Special Jury Award, relied on testimonies from twenty-four military informants inside the army, who disclosed the use of artificial-intelligence algorithms to direct strikes and select targets that caused civilian deaths in the besieged Gaza Strip.

The documentary took its name from a Hebrew military acronym that denotes the expected civilian losses in air raids.It is the term equivalent to collateral damage. The field material relies on accounts from officers and soldiers who refused to reveal their identities, accusing the military establishment of relying on advanced software tools to generate target banks without a genuine accounting of civilian lives, in a war that began on October 7 2023 after an attack led by the Al-Qassam Brigades and Palestinian factions, and has so far resulted in at least 74,000 Palestinian deaths, including more than 21,000 children, according to Gaza’s Ministry of Health statistics.

Netanyahu’s move came via X on Saturday evening, where he directed direct demands to the chief of staff, the head of the Shin Bet security service, and the government’s legal adviser, to launch legal action against those involved in the work, on charges that include serious security violations relating to information secrecy, criminal defamation, and damage to Israel’s diplomatic and security standing. This coincided with Chief of Staff Iyal Zamer’s announcement of investigations into the leak of classified military information used to produce the film, reflecting a shift of the debate from a technical assessment arena to a state security battle.

On the other side, the military establishment denied the informants’ claims last month, describing them as contradictory and illogical, and confirming that target selection and approval remain the exclusive domain of human actors according to operational instructions. Yet the controversy went beyond official data; the film’s directors, Yuffal Abraham and Rachel Zor, faced accusations of treason and threats of stripping citizenship from Netanyahu and Culture Minister Mickey Zohar, while the head of public diplomacy, Tsebi Houtufli, celebrated an oppositional film titled “Naza: The True Story” based on military statements that refute the original narrative, as 43 Israeli human-rights organizations announced their solidarity with the creators of the original documentary and rejected the threats against them.

What this battle reveals does not stop at cinematic documentation; it directly touches the core of technical decisions in our region.In the race among government and security institutions in the Gulf, Egypt and the Levant to embed automated analysis and forecasting models in command centers and sensitive operations, film leaks reveal that merely claiming a “human element in the decision loop” is insufficient to protect systems from slipping. When algorithms generate thousands of recommendations in a few minutes, human review becomes merely a formal procedural signature lacking practical capacity for scrutiny, which requires regional technology managers to rewrite model audit standards and legal accountability chains, rather than relying solely on ready-made black box outputs.

The shift of the accusation from algorithmic accuracy to leak pursuit demonstrates that the sensitive data surrounding decision-making models will remain the most fragile front in any sovereign software architecture. Watch how legal pressure escalates when the tool’s operation method is revealed, and note that AI governance starts with documenting inference logs and assigning criminal liability for each automated decision before the system is deployed in practice.