Deepfakes: Antisemitism’s new calling card is hiding in plain sight

In many ways, the events of Oct. 7 closely mirrored past Palestinian acts of aggression toward the Jewish state. Once again, terrorists masquerading as “freedom fighters” wounded and killed innocent Israelis in a crowded public space. Also once again, Israel’s response has led to the unfortunate deaths of people in Gaza. Once again, anti-Israel zealots, fueled by Hamas propaganda, have united in widespread condemnation and false accusations of “apartheid.”

Sadly, it’s a script that we are all too familiar with here in Israel. And while the scenes of Hamas’ unique brutality on Oct. 7 are seared into the minds of all Israelis, there is one element of this conflict that no one could have foreseen and which poses an even graver long-term danger not only to Israel, but to Jews around the world.

According to a recent World Economic Forum report, “deepfakes” are the world’s biggest short-term threat. The Davos Group contends that digitally manipulated and synthetic media threatens to erode democracy and polarize societies across the globe.

As if on cue, propagandists have been working overtime to exploit this latest technological advancement to push pro-Hamas narratives and inflame public opinion about Israel. Examples of this bad-faith operation have popped up all over the world. 

In January, a Facebook post went viral in India after a user claimed that Israeli Prime Minister Benjamin Netanyahu had declared Jan. 22nd a public holiday. The post, which turned out to be fake, was part of a campaign to incite anger against Israel among India’s nearly 200 million Muslims over the inauguration of a newly constructed Hindu temple, set for the very same day.

Online social media users have also falsely claimed that video footage from the Russia-Ukraine war is of Israeli soldiers systematically destroying Gaza. Even scenes from the popular video game Arma 3 have been presented as actual videos of Israeli troop movements. Other deepfakes show supposed Israeli missile strikes, tanks razing Gazan neighborhoods, and Palestinian families languishing in the rubble.

No example demonstrates this worrying phenomenon more than what occurred on Oct. 17. That day, the New York Times reported that the Israeli military had bombed al-Ahli Arab Hospital in Gaza City. The footage used as evidence in the Times’ report turned out to be a deepfake. Although the paper of record later corrected its story, the clarification was too late to thwart the subsequent anger toward Jews and supporters of Israel.

Compounding the danger that deep fakes pose for Jews is how difficult they are to identify — even expert watchdog groups don’t always know what to look for. While the ADL reported in December a shocking 337 percent increase in antisemitic incidents since Oct. 7, the group could not present physical evidence connecting this rise in anti-Jewish sentiment to any one specific deep fake video.

Any side in a war has the capacity to manipulate its detractors, but there is no denying the particular effectiveness of deepfakes in provoking anger and violence toward Jews. At this very moment, hundreds of antisemitic cohorts around the world are using new tools. Google’s video generation AI STUNet, which can determine with pinpoint accuracy the location and movement of objects in a video, is being used to produce and distribute deceptive digital content at an unprecedented pace. 

Currently, little stands in the way of enterprise-quality deepfakes produced by merchants of lies and deception. Initiatives to identify and label deep fakes such as Meta’s Stable Signature and The Coalition for Content Provenance and Authenticity (C2PA) are welcome efforts. But as researchers at Sumsub — a company that monitors deep fakes — have shown, these programs barely scratch the surface of the problem. 

Where these initiatives have failed the Jewish and pro-Israel communities, Decoding Antisemitism and the Tel Aviv Institute laboratory are providing technical analysis of deepfake media and the interplay between human and algorithmic detections. Through artificial intelligence and qualitative analysis, we gather real-time information to identify anti-Israel and antisemitic deepfakes, which we are then able to report to major news outlets before they inflame tensions by publishing them as authentic news stories.

As effective as these tools and methods have been for sussing out deep fakes, no single strategy has proven more effective than properly educating social media users. Our research provides online influencers with a suite of tools for assessing social characteristics of digital content and mitigating vulnerabilities to deepfake generation technologies.

As much as Israelis want peace with our Palestinian neighbors, we are also acutely aware of the daily threat that we face. And while the ground battle may last for several more months, the war against deep fakes will persist far beyond this current conflict. We encourage the vast and diverse demographic of 200 million Americans with favorable to neutral sentiments about Israel to remain diligent about the threat that deepfakes pose, not only on Israelis of all backgrounds, but Jews worldwide.

Ron Katz is President of the Tel Aviv Institute. Matthias J. Becker, a linguist at the Technical University, Berlin, leads the international research project Decoding Antisemitism and is a visiting fellow at the Tel Aviv Institute.