Israeli insiders have described the targeting and profiling of Palestinians as “a video game” and “insanely fun”
A new documentary alleging that Israel relied on powerful AI tools to kill thousands in Gaza has sent shockwaves through the Venice Film Festival, with one of the film’s insiders describing planning airstrikes as “insanely fun.” The Israel Defense Forces (IDF) has categorically rejected the film’s portrayal, insisting that the whistleblowers cannot be verified.
The 80-minute-long film NAZA, which premiered on Thursday, is built around testimony from 24 anonymized Israeli military and intelligence insiders. Its title alludes to an Israeli military shorthand for expected “collateral casualties” in an attack.
The film, which was made by two Israeli Oscar-winning filmmakers, Yuval Abraham and Rachel Szor, received a roughly 25-minute standing ovation – the longest on record at the event. The footage also showed a full audience, with some viewers holding Palestinian flags.
In the film, Israeli insiders claim that the IDF used an AI system known as ‘Lavender’ to identify people for assassination. The filmmakers’ sources also alleged that the tool operated based on thousands of hacked phones, with even a single phrase, “daddy’s home,” picked up on a child’s phone being enough to launch an attack.
Other sources told the tale of brazen disregard for humanitarian laws, and a “general atmosphere” of “hatred” towards Palestinians. In one of the most chilling revelations, an insider claimed that at one point a strike on a single Hamas member was approved despite an expected toll of 500 civilians.
“It was done on a mass scale, like a factory”, one source said, with others describing the profiling and targeting as “a video game” and “insanely fun.”
“We are letting the soldiers and the officers speak about what they did,” Abraham told AFP, saying the filmmakers wanted to document in detail “how the systems of killing in Gaza looked” and operated.
On Friday, the IDF vehemently rejected the allegations, arguing that the identities of the whistleblowers “cannot be verified,” and that it is impossible to establish whether they served in the military at all.
It also stressed that allegations pertained to “national and military strategy that are clearly beyond the knowledge of junior-level soldiers.” The IDF added that targeting decisions are made “by human personnel only,” and that its operations are only directed at “terrorist organizations.”
The IDF has itself acknowledged using AI for military purposes, including a system known as ‘The Gospel’, to rapidly produce targeting recommendations.
In April 2024, several media outlets reported, citing Israeli intelligence sources, that Lavender had at one stage marked about 37,000 people as suspected terrorists, with the personnel sometimes spending only seconds reviewing machine-generated recommendations. Some sources quoted by The Guardian also alleged that “during the early weeks of the war they were permitted to kill 15 or 20 civilians during airstrikes on low-ranking militants.” The IDF has rejected the allegations.
