Deepfakes, bots and distrust: How AI is shaping Israel's election

In a 19-second clip

posted to X, Gadi Eisenkot, a candidate for prime minister, stands with his arms outstretched and a huge smile on his face. In slow motion, he runs through a meadow toward a person who resembles his son, wearing a shirt with the words "Israeli Unity." But Eisenkot sails past him and instead embraces Mansour Abbas, a prominent Arab politician who leads the  Ra'am party, running in the elections.

The video was posted on Prime Minister Benjamin Netanyahu's account. Opinion polls

show Netanyahu and Eisenkot, who also happens to be a former chief of staff of the Israel Defense Forces, in a tight race. The AI-generated video appears to be designed to play on anti-Arab sentiment — a big part of Netanyahu's campaign. 

The clip was viewed more than 1 million times and stirred outrage. Eisenkot's son was killed in 2023 during the Israel-Hamas war.

According to Ze Be AI, a research project of the Open University of Israel, monitoring AI content in the 2026 legislative election campaign, this clip is just one of more than 1,800 posts with AI-generated content that Netanyahu's Likud party had released

 throughout the election campaign by mid-October. That represents about a quarter of the Likud party social media materials analyzed by Ze Be AI.   

On October 27, Israelis will vote in an election that could oust Netanyahu from power or cement his grip on the country for the coming years. Political scientists are already calling it Israel's first AI election. Parties have been deploying AI across every aspect of campaigning — from creating outrageous videos aimed at stirring controversy to robocalling voters and unleashing bots online.

"There is a scenario where, in a strategic point shortly before the election, AI could or would be used in ways that would steer controversy that could tilt the election results," said Anat Ben-David, a media researcher leading the Ze Be AI project at the Open University of Israel. 

AI videos go mainstream

Ben-David and her research team have been tracking

 all of the content posted on social media by politicians since the end of July. By mid-October, they had collected more than 44,000 posts across Facebook, Instagram, X and TikTok and found that over 14% of the material used AI in some way. 

A hybrid approach was most common, with AI being deployed for a part or segments of an ad: in images, music, subtitles, logos or full-fledged deepfakes.

Some fringe parties use the greatest ratio of AI in their content. But Likud is the winner by far when it comes to sheer numbers of AI use in content analyzed by the team. Netanyahu's party is followed his left-wing rivals, the Democrats, and then Eisenkot's party, Yashar — though at far lower numbers. 

Israel's Central Elections Committee issued a new rule

in July mandating that all AI-generated or -manipulated content in election campaigning be labeled. Beyond that, there are few limits. 

"There's no AI Act, no Digital Services Act, and the privacy law is very outdated," Ben-David said. "Then you see new [AI] uses that would be impossible to see in Europe."  

Some of those uses have come from Israel's most prominent politicians.  

Yair Golan, head of the Democrats, has been trying to capitalize on popular internet trends. One video

labeled as AI content features him sitting under a ceiling fan — with each rotation of the blade, his outfit changes: from an army uniform to a T-shirt with the popular slogan "bring them home now" (referring to Israeli hostages in Gaza) to a suit and tie. 

In one photorealistic AI video, posted

 by Finance Minister Bezalel Smotrich in September, Israel's attorney general and Supreme Court justices are apprehended at an anti-government protest. Attorney General Gali Baharav-Miara has frequently clashed with the ruling coalition. 

And National Security Minister Itamar Ben-Gvir shared an AI-generated video

  of Palestinians being wheeled on a conveyor belt into what looks like a concentration camp. That post drew widespread condemnation, with Germany's Foreign Ministry calling it

"completely unacceptable" and "abhorrent." 

The election authority has shown some willingness to intervene. In July, its chairman ordered

Smotrich to take down an unlabeled AI video it called misleading.  

A new generation of robocall

It's not just deepfakes and ads: AI is also being used to reach voters directly. 

Some voters are getting a new kind of AI-powered robocall — far more conversational than the robotic automated calls of the past. Nitsan Yasur, a disinformation researcher at the Israel InternetAssociation, shared a recording with DW of a four-minute phone call.

In it, a voice introduces itself as a pollster named Danielle with an Israeli polling institute and asks for his voting preferences. When Yasur asks if the voice is human, it admits, "there is no human agent on the line," before continuing to ask, "would you like to continue with the short survey?" 

The bot fields several of Yasur's questions, while always steering back to who he would vote for. But when Yasur asks five times who commissioned the survey, it doesn't reveal the information. So it remains unclear whether this was part of an election campaign or a poll. What is clear is how advanced robocalling has become. 

"You don't need real people making real phone calls, which are obviously limited in time and capacity," Tehilla Shwartz Altshuler, a senior fellow at the Israel Democracy Institute, told DW.  

Efforts to shape what chatbots tell voters

Some Israelis are also turning to AI for advice on how to vote.  

About a quarter of Israelis said they would consider using AI tools or other online services to inform their voting decisions, the Israel Internet Association found in a survey

.

That raises concerns about generative engine optimization (GEO): creating online content designed to be picked up by AI chatbots and shape their answers. The Israeli government has already tapped this industry

. According to filings under the Foreign Agents Registration Act, it paid a US firm to build websites meant to feed LLM systems with content supportive of the government in order to make AI chatbots more pro-Israel. 

"The same industries are going to be used now inwards in order to affect the elections," she said. 

Bot accounts promote Netanyahu online

There's fake content, and then there are fake audiences. Scooper, an Israeli company specializing in media monitoring, told DW that it discovered a network of inauthentic accounts boosting Netanyahu on social media. On X, more than 1,000 new accounts have been appearing daily and immediately following the Israeli prime minister. His Instagram following has also surged in recent weeks.  

About 70% of the accounts that share Netanyahu's posts on X are based outside of Israel and resemble bots, Scooper research manager Gil Feldman told DW. Feldman said the company flagged them by looking for suspicious behavior such as round-the-clock posting, recent creation dates and missing profile pictures. 

And it's not just Netanyahu: The Israeli watchdog FakeReporter found a bot campaign

linked to Ben-Gvir's campaign, which his office denied. FakeReporter discovered a network of more than 100 accounts, some using AI-generated images and identities, targeting Smotrich and others.

Though bot networks are nothing new, AI can make them cheaper and easier to run. And, because Israelis are early and eager adopters of new technology, Shwartz Altshuler sees the country as a test case for how AI is reshaping elections around the world. 

"The whole machinery of political propaganda is becoming totally different," Shwartz Altshuler said. 

Edited by: Uta Steinwehr, Chiponda Chimbelu, Joscha Weber, Shani Rozanes