In the weeks since Hamas launched its October 7 surprise attack on Israel, the ensuing conflict has generated an unprecedented wave of disinformation, an “algorithmically driven fog of war” that has tripped up major new organizations and left social media companies floundering.
Yet, amid all of the deceptive images and video moving around on social media, the content generated by artificial intelligence tools has remained relatively peripheral. Even as some wondered if the Israel-Hamas war would be the first conflict dominated by false generative AI images, the technology has had a more complex and subtle impact.
“There are definitely AI images circulating but not to the degree where I think it’s playing a central role in the spread of information,” says Layla Mashkoor, an associate editor at the Atlantic Council’s Digital Forensic Research Lab, which studies online disinformation.
Primarily, Mashkoor says, AI-generated disinformation is being used by activists to solicit support—or give the impression of wider support—for a particular side. Examples include an AI-generated billboard in Tel Aviv championing Israel Defense Forces, an Israeli account sharing fake images of people cheering for the IDF, an Israeli influencer using AI to generate condemnations of Hamas, and AI images portraying victims of Israel’s bombardment of Gaza.
“In terms of general use that I’ve been seeing online, it’s mostly been to drum up support, which is not among the most malicious ways to utilize AI right now,” she says.
A key factor here is the sheer amount of misinformation circulating, which makes it difficult for AI images to shape conversation. “The information space is already being flooded with real and authentic images and footage,” says Mashkoor, “and that in itself is flooding the social media platforms.”
This point is reflected in a recent paper from the Harvard Kennedy School Misinformation Review about the role generative AI might play in the spread of false info worldwide. In it, the authors write that concerns about the effects of the technology are “overblown.” While, yes, gen AI theoretically lets people proliferate misinformation at a futuristic rate, those who seek out this misinformation—often those who have “low trust in institutions … [or are] strong partisans”—already have a surfeit of familiar nonsense to pursue, from conspiracy theory websites to 4chan forums. There is no demand for more.
“Given the creativity humans have showcased throughout history to make up (false) stories and the freedom that humans already have to create and spread misinformation across the world, it is unlikely that a large part of the population is looking for misinformation they cannot find online or offline,” the paper concludes. Moreover, misinformation only gains power when people see it, and considering the time people have for viral content is finite, the impact is negligible.
As for the images that might find their way into mainstream feeds, the authors note that while generative AI can theoretically render highly personalized, highly realistic content, so can Photoshop or video editing software. Changing the date on a grainy cell phone video could prove just as effective. Journalists and fact checkers struggle less with deepfakes than they do with out-of-context images or those crudely manipulated into something they’re not, like video game footage presented as a Hamas attack.
In that sense, excessive focus on a flashy new tech is often a red herring. “Being realistic is not always what people look for or what is needed to be viral on the internet,” adds Sacha Altay, a coauthor on the paper and a postdoctoral research fellow whose current field involves misinformation, trust, and social media at the University of Zurich’s Digital Democracy Lab.
That’s also true on the supply side, explains Mashkoor; invention is not implementation. “There’s a lot of ways to manipulate the conversation or manipulate the online information space,” she says. “And there are things that are sometimes a lower lift or easier to do that might not require access to a specific technology, even though AI-generating software is easy to access at the moment, there are definitely easier ways to manipulate something if you’re looking for it.”
Felix Simon, another one of the authors on the Kennedy School paper and a doctoral student at the Oxford Internet Institute, cautions that his team’s commentary is not seeking to end the debate over possible harms, but is instead an attempt to push back on claims gen AI will trigger “a truth armageddon.” These kinds of panics often accompany new technologies.
Setting aside the apocalyptic view, it’s easier to study how generative AI has actually slotted into the existing disinformation ecosystem. It is, for example, far more prevalent than it was at the outset of the Russian invasion of Ukraine, argues Hany Farid, a professor at the UC Berkeley School of Information.
Farid characterizes the technology as a “specter” hanging over audio and video that purports to come from the conflict, and that he receives half a dozen to a dozen calls daily from reporters enquiring about veracity. “For a lot of people, the ability to dismiss inconvenient facts is absolutely playing a role in this conflict,” he says.
Farid cites multiple examples that immediately drew this kind of dismissal, including people pointing to various pieces of digital evidence about who was behind the missile strike on the Al-Ahli Arab Hospital in Gaza, as well as images of children buried under rubble, some real, some fake.
Some of the most prominent examples of this are the photos of burned children Israeli prime minister Benjamin Netanyahu posted on his X account. After they were published, someone fed the images into the detection tool AI or Not, which the users said determined that they were AI. Farid says his team analyzed the photos and concluded AI wasn't used, but the seed of suspicion had already been planted. Things were further confused when someone used AI to replace the child in one of the images with a puppy.
“And then that went online, and then people are like, ‘Well, wait a minute, if he could have made that one, then that one could be fake, and now these two versions are going around, and everybody started saying, ‘Oh, the puppy was the original, this is the fake,’ and then you just muddy the water,” Farid says.
In other words, this distribution follows a historic pattern: Misinformation gets shared on social media, then amplified via algorithms and humans. “In the broader picture, in our ability to reason about a fast-moving highly impactful world, I think this conflict is worse than what we’ve seen in the past,” says Farid. “And I think gen AI is part of that, but it is not exclusively gen AI. That’s too simplistic.”
