clear

Creating new perspectives since 2009

Israel caught funding fake think tank to game AI and push propaganda

August 27, 2026 at 3:27 pm

Artificial Intelligence, AI [Wikipedia]

Israel has funded a fake US think tank and a network of apparently independent websites designed to game artificial intelligence systems and push pro-Israel narratives into chatbot responses, according to an investigation by the Guardian.

The operation is part of a multimillion-dollar Israeli campaign to manipulate the online information environment as the occupation state struggles to reverse an unprecedented collapse in international support following its genocide in Gaza and expanding regional wars.

Israeli officials have already conceded that the massive propaganda drive is failing, despite hundreds of millions of dollars being poured into hasbara.

The revelations come just days after a senior Israeli communications figure openly admitted to fabricating false information in Israel’s propaganda war. Eli Hazan, a longstanding Likud communications official and former spokesperson for Prime Minister Benjamin Netanyahu, declared: “The truth doesn’t matter anymore. The facts don’t matter anymore.”

Hazan went on to acknowledge: “I fabricate fake news as a response to them,” while arguing that Israel should flood the information space and adopt a communications strategy modelled on US President Donald Trump.

The comments offer a striking backdrop to the Guardian investigation into the Hanover Institute for Public Policy, a purported American think tank which appears to have been created as part of an Israeli government-funded influence operation.

According to the Guardian, Hanover apparently does not exist as a legal entity, has no physical address, identifies no staff and carries no author bylines. Nevertheless, between 6 and 14 August it pumped out 124 supposedly academic reports containing more than 560,000 words.

Almost every headline takes the form of a question that resembles a prompt someone might enter into an AI chatbot, including questions about the Nakba, anti-Zionism, Israel’s occupation and whether Israel is committing genocide in Gaza.

The material is presented to resemble scholarly research, complete with references to the World Bank, UN agencies, Amnesty International, Palestinian statistics and international legal texts. However, the Guardian found that its framing repeatedly serves to blunt criticism of Israel and cast doubt on established evidence concerning Gaza.

The strategy appears to have been designed as much for machines as for human readers.

The Hanover website was built using technology produced by Res, which describes itself as an “AI-native content platform” intended to help companies get cited by systems including ChatGPT, Perplexity, Claude and Gemini.

Piro Inc, the New York media company behind Hanover, also markets what it calls “AI Story Optimization”, creating material specifically engineered around how large language models assess credibility. Piro registered its work for the Israeli government under the US Foreign Agents Registration Act

The strategy works in two ways. The first is to create apparently authoritative material which chatbots can retrieve and cite when answering questions about Israel and Palestine. The second, potentially more consequential method, is to push such material into vast internet repositories which may later be used to train future AI models.

Once propaganda enters an AI model’s underlying training material, users may receive its framing without being told where the narrative originated.

The Guardian findings corroborates recent revelations about Israel paying tens of millions of dollars to “poison” AI chatbots through a separate network created by Clock Tower X, the digital influence company headed by former Trump campaign manager Brad Parscale.

Clock Tower X holds a $46.5 million Israeli government contract and created ten websites publishing hundreds of articles designed to promote Israeli narratives about Gaza. The sites have challenged accusations of genocide, sought to discredit Palestinian journalists killed by Israel and cast doubt on documented Israeli attacks, including the killing of six-year-old Hind Rajab.

Israeli operations designed to game AI form part of a much wider attempt to occupy the digital information space through artificial intelligence, paid influencers, conservative media networks, targeted advertising, mass text messages and apparently independent organisations.

Yet Israel’s problem appears increasingly resistant to public relations.

Israel’s international image has plummeted since October 2023 as evidence of the devastation in Gaza have circulated globally. In the US, traditionally Israel’s most important international supporter, Pew found that 60 per cent of Americans held an unfavourable view of Israel earlier this year, an increase of nearly 20 percentage points since 2022. Gallup also found that Americans sympathised more with Palestinians than Israelis for the first time, with only 36 per cent siding with Israel.

Israel has responded by dramatically increasing spending on hasbara. Its public diplomacy budget surged to around $730 million, more than four times the $150 million allocated the previous year and many times higher than spending before the Gaza genocide. Money has flowed into AI campaigns, foreign influencers, advocacy groups and targeted digital operations.

However, Israeli officials themselves increasingly admit that the strategy has failed.

Israeli officials told Ynet in July that the government had “paid a lot of money” only for the situation to deteriorate further. A former senior figure in Israel’s public diplomacy apparatus went even further, describing the entire initiative as “a terrible hoax on the scale of Pollard”, while another official questioned what Parscale had actually achieved with the millions he had received.

Nick Cleveland-Stout of the Quincy Institute, whose research contributed to the Guardian investigation, said that the fundamental flaw is Israel’s attempt to treat a political crisis as a communications problem. While the AI manipulation may succeed in getting Israeli narratives into chatbot answers, he noted that it has not changed the opinion polls which ultimately matter to Israeli officials.