Summary: "NAZA," a documentary alleging intentional large-scale killings by the IDF in Gaza, draws heavily on prior investigations by co-director Yuval Abraham. Those investigations used a small pool of anonymous sources and did not fully trace how AI-generated target recommendations were handled or cross-check claims against military records. Independent reporting — notably by The New York Times — and analysis from West Point's Lieber Institute suggest the tools described are decision-support systems subject to human approval, and that safeguards were modified but not abandoned. Given gaps in verification and limited sourcing, the film's broadest claims warrant careful scrutiny.
“NAZA” Under the Microscope: Why Its Allegations Against the IDF Require Stronger Evidence
Yuval Abraham and Rachel Szor pose with the Special Jury Prize Award for "Naza" during the 83rd Venice International Film Festival on Sept. 12, 2026 in Venice. (Credit: Andreas Rentz/Getty Images)
"NAZA," which won the Special Jury Prize at the Venice Film Festival, presents stark allegations that the Israel Defense Forces (IDF) conducted a campaign of deliberate, large-scale killings in Gaza. Such grave charges demand careful, verifiable evidence. A close look at the reporting that underpins the documentary — much of it produced by co-director Yuval Abraham — suggests significant gaps in corroboration and source depth.
Background And Source Issues
Because "NAZA" has not been widely released and this writer has not viewed the full film, much of its most disturbing material has already circulated in earlier investigations — including pieces published in The Guardian, which is listed as a producer on the documentary. Crucially, Abraham himself produced several of those antecedent reports.
Those reports rely heavily on a small pool of anonymous intelligence sources. The film reportedly draws on roughly two dozen intelligence-corps interviewees, many of whom had already appeared in Abraham's earlier stories. That limited and overlapping source base makes independent corroboration especially important for the film's most sweeping claims.
What The Reporting Says About AI Tools
Abraham's reporting described three nicknamed systems: "Gospel," an AI tool said to generate target recommendations; "Lavender," a ranking system for suspected militants; and a third system dubbed "Where's Daddy?," reported to flag suspected militants' presence in homes. Some interviewees described these systems in alarmist terms — for example, calling "Gospel" a "mass assassination factory."
However, published accounts and independent experts indicate an important distinction: these systems appear to function as decision-support tools, not as autonomous weapons. Analysts at West Point's Lieber Institute have confirmed that "Gospel" and "Lavender" are intended to aid analysts by identifying potential targets; human decision-makers are reported to remain responsible for assessing proportionality and authorizing strikes.
Corroboration And Alternative Reporting
Abraham's pieces did not clearly trace what happened to AI-generated recommendations after they left the system, nor did they systematically cross-check interview claims against military orders, specific strike records, or satellite imagery. In contrast, The New York Times' December 2024 investigation interviewed more than 100 soldiers and officials — including more than 25 people directly involved in the vetting and strike chain — and reviewed military records and imagery.
The Times reported that in the immediate aftermath of Oct. 7 the IDF briefly raised permitted collateral figures for certain preplanned strikes — for example, increasing some ceilings from zero-or-five civilians to as high as 20 — and that these allowances were subsequently tightened (reduced to 10 on Nov. 5, 2023, and later subject to special approval). The Times also found no documented strike in which 500 civilians were killed, and the IDF stated no single strike on that scale was ever planned or approved. These distinctions matter when judging claims that safeguards were broadly or permanently abandoned.
Questions About Casualty Figures
Abraham's later claim that 83% of fatalities were civilians has been widely questioned. The IDF has said it could identify 17% of fatalities by name as combatants; Abraham's computation that the remaining 83% were civilians overlooks unidentified fighters who would not appear on that list. In active conflict zones it is often impossible for any army to name every combatant killed, so drawing definitive civilian-versus-combatant ratios from incomplete lists is methodologically fraught.
Conclusion: Demand Verification
"NAZA" promises to "describe things as they really are," but the primary investigations underpinning its allegations rely on limited anonymous sourcing and leave important evidentiary questions unanswered. Independent expert analysis and more extensive documentary corroboration — military orders, strike logs, forensic records and imagery — are needed before accepting the film's most sweeping conclusions. Viewers should approach those claims with healthy skepticism and insist on transparent verification.
Shlomi Ben-Meiris
Hebrew Editor, Committee For Accuracy In Middle East Reporting And Analysis (CAMERA)
Help us improve.
