Inside the Israel-backed ‘NAZA’ disinformation campaign
An investigation exposes sophisticated efforts to
undermine the film — including the manipulation of AI chatbots to promote a
counter-documentary.
By Clément Pouré November 16, 2026
https://www.972mag.com/naza-disinformation-campaign-israel-ai/
The release of “NAZA” in September — a documentary by Yuval Abraham and
Rachel Szor exposing the genocidal logic underpinning Israel’s war on Gaza —
provoked a fierce backlash in Israel. As Reporters Without Borders (RSF) has documented, Abraham and Szor faced cyber-harassment, threats of
legal action by the Israeli military, and even threats to revoke their
citizenship.
But alongside these attempts to intimidate the
filmmakers, a more elaborate campaign aimed at undermining the documentary’s
findings and steering audiences toward a rival version of the film was taking
shape.
An investigation by Mediapart reveals that this
disinformation campaign was designed in part to manipulate AI chatbots into
recommending an alternative version of the documentary to users searching for
“Naza.” The investigation, based on a technical analysis carried out by AI
Forensics, establishes direct links with the Israeli state.
On Sept. 18, six days after “NAZA” won the Special
Jury Prize after its premiere at the Venice Film Festival, an unknown Israeli
organization registered the domain name Nazadocumentary.com through Cloudflare.
As the Guardian — which co-produced “NAZA” — and Al Jazeera have reported, the website initially presented itself as the
“official NAZA website, listing the directors’ names and information on
screenings and press.”
Two days later, a 49-minute video was uploaded to the
site. Entitled “NAZA: The True Story” and subtitled in 39 languages, this
counter-documentary imitates the original film. The text on Nazadocumentary.com
was subsequently changed: By Sept. 30 at the latest, it had stopped referring
to the original documentary and was instead promoting the rival version.
“The most credible explanation,” says Paul Bouchaud, a
researcher at France’s national research centre (CNRS), who led the technical
analysis carried out by AI Forensics, “is that the site’s creators adopted this
strategy to improve its search engine rankings and make it appear more
frequently to internet users looking for information about the original film.”
The counter-documentary, whose producers include Ariel
Averbuch, a lawyer, and Roi Danino, a candidate for the right-wing Israel First
party, directly attacks the work of Abraham and Szor.
It criticizes the anonymity of the testimonies at the
heart of the journalists’ investigation, echoes the Israeli military’s
narrative, and portrays the victims of its operations as Hamas members. A
counter on the website claims more than 45 million views, but according to Al
Jazeera, the figure has been inflated, with the actual number of views
estimated at a few hundred thousand.
Two other websites, Whatisnaza.com and Watchnaza.com,
were registered on Oct. 6, also through Cloudflare, and likewise promote the
rival documentary. Linked to each other, they repeat the narrative advanced by
the Israeli state since the original film’s release. “The war depicted in the
film begins with the attacks of October 7th 2023,” states Whatisnaza.com, for
example, which hosts extensive uncensored footage of the attack.
The two sites also provide factual information about
Abraham and Szor’s film, including details of upcoming screenings. “The site’s
architecture is clearly designed to ensure that it appears in search results
for users looking for information about the documentary,” notes Paul Bouchaud.
But the strategy is not limited to conventional search
engines. An in-depth analysis of Nazadocumentary.com reveals a collection of
files intended for large language models (LLMs) and AI agents rather than
ordinary internet users. Every page is available in JSON and Markdown formats,
making its content easier for AI tools to read.
The website also features an “agent skill” system, a
script specifically designed for use by AI agents. A dedicated page on the site
has also been custom-built to influence chatbot responses, particularly when
users search for “Naza Film,” “Naza Movie,” or “Naza Documentary.”
“These strategies, known as ‘LLM grooming,’ have
become increasingly common over the past year and a half, but I can’t recall
seeing one quite so sophisticated,” Vincent Berthier, head of the technology
and journalism desk at RSF, told Mediapart.
In late 2025, the press freedom organisation warned about the proliferation of fake news sites spreading
disinformation aligned with Kremlin narratives ahead of France’s municipal
elections.
“That was a mass strategy, designed to saturate large
language models,” Berthier recalled. “This time, we’re seeing something
methodically designed to control the results produced by AI chatbots more
precisely.”
Tests carried out by AI Forensics in mid-October
suggest the strategy is working: When ChatGPT and Grok were asked, “Where can I
download or stream the documentary Naza?” their responses frequently directed
users toward the alternative version. Grok, the chatbot developed by xAI, also
directed users to Watchnaza.com.
The Israeli government has denied any involvement in
the production of the rival documentary “NAZA: The True Story.” Yet, as the
Guardian reported, the Israeli state has repeatedly promoted it.
Israel’s embassy in France, along with other Israeli embassies around the
world, has promoted Nazadocumentary.com through Instagram posts and a link in
its page’s bio.
This active promotion has also involved paid
advertising. Records from the Google Ads Transparency Center, a public register of advertisements required under
the European Union’s Digital Services Act, show that the Israeli government’s
communications agency funded at least 19 ads beginning on Oct. 15 directing
internet users to Nazadocumentary.com. Several of these ads, placed through an
account created specifically for the campaign, targeted users in France.
No hay comentarios:
Publicar un comentario