Generative Artificial Intelligence (GAI) and the Israel-Hamas war

Promoters of conspiracy theories and hate are using generative artificial intelligence (GAI) to create misleading content about the Israel-Hamas war. This not only amplifies confusion and hate on social media; it can also cause some to doubt the validity of actual war images, creating unnecessary suspicion at a time of deeply polarized public opinion. ADL’s previous research includes recommendations on what tech companies should do to address heinous abuses of GAI technology such as those listed below. 
 
Misleading GAI images 

Disinformation peddlers are using GAI technology to tell a fictional story about the conflict. Many of these images have obvious, tell-tale signs of being inauthentic, but can fool the untrained eye and may be difficult to distinguish from the many photos showing actual victims. 

On X (formerly Twitter), some GAI images depict what looks like injured or frightened children running from air strikes in Gaza. One post on X includes a series of them with the hashtag #StopIsraelCrimes.

Another image shows a fake AI-generated “tent city,” which users claim was constructed for Israeli refugees. When it first started circulating, the image was popular on X and Instagram, where anti-Israel commenters hoped the camp would be destroyed. Strange artifacts in the image, such as an Israeli flag with two Stars of David on it, hint that the content is inauthentic. 

Deepfake videos, which generally require several GAI tools to create, have also gained traction by presenting fictionalized “commentary” on the war from public figures. While these videos are often advertised as deliberately satirical, they can easily be shared without that context, creating further confusion. 

For example, a deepfake video of model Bella Hadid, who is publicly critical of Israel, shows her retracting her prior comments about Palestine and expressing support for Israel.

In reality, the video was likely created using synthetic speech, which uses AI to mimic real voices that have been “cloned” from audio samples of authentic speech. The synthetic speech was then mapped onto old footage of Hadid speaking at the Global Lyme Alliance in 2016.  

Offensive and violent GAI propaganda 

Racist and antisemitic Telegram and 4chan users are creating hateful and misleading propaganda about the war using GAI tools. 

Some channels and message boards are dedicated solely to crowdsourcing memes, encouraging “memetic warfare” and advising users to create "propaganda for fun.” Posts provide instructions on how to use GAI tools to produce radicalizing images based on text prompts. Many of these images reference the Israel-Hamas war -- mostly glorifying Hamas, demonizing Israel and spreading already-debunked false narratives. 

Several memes depict the moment Hamas paragliders crossed into Israel on October 7. One Telegram post includes the caption, “Getting the view before tearing up the Torah.”

Another GAI image on 4chan and Telegram shows paragliders over a burning building. At a distance, the meme reveals an image of Adolf Hitler.

An image shared on an X account dedicated to GAI memes shows a visibly Jewish man standing next to a cartoon bomb in a hospital. This is one of many recent GAI memes depicting the explosion at Al-Ahli hospital in Gaza on October 17, which many initially blamed on Israeli airstrikes. This claim has been widely debunked via multiple analyses from independent experts, the media, the U.S. government and others, which point to the explosion being caused by a rocket misfire from within Gaza.

A meme shared in an AI “art” Telegram channel shows an approaching military tank with a large nose, along with the caption, “BREAKING: Israeli tanks have been spotted moving to Gaza.” The image is clearly drawing inspiration from a classic hateful trope about Jewish people having disproportionately large noses.

GAI encourages doubt of actual documented violence 

Perhaps the most worrying side effect of GAI is that its increasing popularity on social media has sowed doubt over real images of graphic or traumatic content. This phenomenon, often referred to as the “liar’s dividend,” has directly impacted the discourse surrounding the war in Israel and Palestine, bringing the authenticity of reported violence into question. 

On X, 4chan and Telegram, many claimed that an image of a burnt body believed to be an Israeli victim was run through “AI-detector tools” and deemed fake. According to experts, including our own analysts, such tools are unreliable and routinely produce inconsistent results.

Trolls on 4chan promoted the theory that the original photo showed a puppy from an animal rescue wrapped in a towel, and that propagandists used AI tools to edit the puppy out and put a burnt body in its place. However, the puppy photo in question has not been traced to any legitimate source online beyond the 4chan edit. ADL analysts have independently verified images of other similarly burned victims in Israel. 

Photos from Israeli homes showing the bloody aftermath of Hamas’s October 7 attack have also been deemed fake by anti-Israel users online. One image on X alleges that the blood in the photo has been “staged” and that the knife looks like it was generated by AI.

On X, a spokesperson for the Israeli government shared photos believed to show human remains at Kibbutz Be’eri, including teeth. Users on both X and Reddit declared without evidence that the images were “fake AI,” alleging that the photos were a product of “Israeli propaganda” meant to fool the masses.