Israel funded fake thinktank that used AI to generate pro-Israel propaganda for chatbots, filings show

A fake thinktank funded by the Israeli government has published more than half a million words of pro-Israel propaganda in nine days, using a commercial platform designed to make its content attractive to AI chatbots, according to disclosures filed with the U.S. Department of Justice.

The site, branded as the Hanover Institute for Public Policy, has no physical address, no named staff, and no bylines. Its terms of service are governed by the laws of an unnamed state. The New York media production company Piro Inc registered the site under the Foreign Agents Registration Act (FARA) on August 14, 2026, as material it distributes for the Israeli government, The Guardian reported.

The institute’s reports present Israel’s positions on contested topics — torture of Palestinian prisoners, war crimes allegations, and whether Israel deliberately starved Gaza’s population — as neutral research. The content is built on a platform that promises to optimize text so that large language models will cite it in answers, effectively gaming the training data of chatbots.

Piro Inc’s FARA filing is one part of a broader effort financed by tens of millions of dollars from the Israeli government, routed through third parties including the European advertising group Havas Media. A contract between Havas and another American firm working on the campaign commits to the “deployment of websites and content to deliver GPT framing results on GPT conversations,” according to The Guardian.

The Hanover Institute’s existence was first reported on August 14. Since then, it added a funding page that states: “An earlier version of this page said the Institute operated without external funding, and that was true when it was written.” The Guardian contacted Piro Inc, Havas Media, and the Israeli government advertising agency LaPam for comment; none have responded publicly.

The AI-powered propaganda campaign coincides with concrete actions on the ground in the occupied West Bank and East Jerusalem that further erode Palestinian civilian infrastructure.

On August 4, Israeli forces demolished two administrative rooms and a storage room at the Shaab al-Butum school in Masafer Yatta, a village in Area C of the occupied West Bank, where Palestinian building permits are nearly impossible to obtain. The rest of the school is also threatened with demolition, with a court case scheduled for September 15, Al Jazeera reported. Children returning to school in September face an uncertain future, with students like 17-year-old Farah telling Al Jazeera she fears the complete loss of the only school she has ever attended.

Meanwhile, Israel seized an UNRWA training centre in East Jerusalem. UNRWA chief Christian Saunders told Al Jazeera the seizure is part of Israel’s efforts to eliminate the agency and the Palestinian refugee issue, adding that UNRWA is “not planning on going anywhere,” as reported.

For the tech community, the implications are direct. The Hanover Institute case shows that foreign governments are now systematically funding content farms to manipulate the training data of commercial AI models. Engineers building chatbots, search engines, or retrieval-augmented generation systems need to verify the provenance of sources that appear authoritative. A thinktank with no staff, no address, and a FARA filing is a red flag — but many similar operations may not be registered. The Israeli campaign, with its Havas contract explicitly targeting GPT framing, indicates that chatbot citation is now a deliberate target for propaganda operations.

No independent verification of the Hanover Institute’s claims has been found. The site continues to publish, and its content may already be influencing AI responses on topics of international law and human rights.