Israel’s New Hasbara Target Isn’t You. It’s ChatGPT.
A seemingly independent think tank published hundreds of reports designed to put Israel’s arguments into the AI information ecosystem
Illustration: Priscilla Du Preez via unsplash
Israel’s public diplomacy campaign appears to be moving beyond social media and traditional news outlets — and toward a new audience: artificial intelligence systems.
An investigation by The Guardian found that the Hanover Institute for Public Policy, a website presenting itself as a research organization, published 124 reports totaling more than 560,000 words in just nine days in August. As of this morning, the website is still online, but it has not published a new report since August 12.
The reports address some of the most contentious questions surrounding Israel, including allegations of genocide, war crimes, torture and the displacement of Palestinians.
The unusual part is not simply the volume. It is the way the material appears to have been produced and optimized.
The institute has no apparent legal existence, physical address, named researchers or bylines. Its reports are presented in a quasi-academic format and often take the form of questions users might ask an AI chatbot: “Is Israel Carrying Out a Genocide in Gaza?” or “Did Israel Expel Palestinians From Their Land?”
According to The Guardian, the site was created as part of an Israel-backed campaign and registered under the U.S. Foreign Agents Registration Act by New York-based media company Piro Inc.
Its technical infrastructure offers an even clearer indication of the intended audience. The site initially carried an llms.txt file associated with Res, a company that markets technology designed to help websites get cited by ChatGPT, Claude, Gemini and Perplexity.
This is the emerging world of generative engine optimization: trying to influence not what appears in a Google search, but what an AI system tells a user.
The strategy also raises an uncomfortable question. If governments can flood the web with carefully framed, machine-readable material, AI systems may increasingly become another arena for influence campaigns — one where the audience cannot always see who is trying to influence the answer.
The next battle for public opinion may not be over what people read.
It may be over what machines learn to tell them.