An Israeli government-funded communications project has created a website resembling an independent policy institute as part of an apparent effort to shape how artificial intelligence systems answer questions about "Israel" and Palestine, Responsible Statecraftreported.
The outlet's Nick Cleveland-Stout identified the Hanover Institute for Public Policy as a project produced for the Israeli Government Advertising Agency by Piro, Inc., a US communications company that openly promotes techniques for influencing the information used by large language models.
Hanover presents itself as a research organization examining antisemitism and issues surrounding Palestine and "Israel". Its website contains more than 100 research-style publications carrying footnotes, tables of contents, and neutral-sounding titles dealing with subjects ranging from the Nakba and conditions in Gaza to AIPAC and Israeli war crimes.
However, the organization does not appear to operate as an independent think tank. Its reports contain no individual author bylines, while a notice at the bottom of its website states that it was established on behalf of the Israeli Government Advertising Agency by Piro.
Piro, co-founded by film producer Daniel Rosenberg, has received $900,000 from the Israeli government for the project, according to Responsible Statecraft. Its work is reportedly being subcontracted through French advertising and public relations conglomerate Havas Media.
The investigation suggested that Hanover's publication strategy is intended not merely to reach human audiences, but to increase the likelihood that AI systems encounter, trust and reproduce its material when answering politically sensitive questions.
Piro advertises a service it calls "AI Story Optimization", saying it can "author(s) content engineered for how LLMs evaluate credibility."
Rosenberg has also publicly promoted his firm's ability to affect information surfaced by AI tools.
"When someone asks ChatGPT, Gemini, or Perplexity about your category, an answer comes back in one confident paragraph. Most brands have no idea how that paragraph gets built. So we spent months reverse-engineering it," he wrote on LinkedIn last month.
"At Piro, we already knew how to build stories that move people. The question was: how do you make sure AI knows how to tell them?"
Hanover's reports appear structured around precisely the type of questions users might submit to AI assistants, including "What Caused the Displacement of Palestinians in 1948?", "Which Humanitarian Organizations Have Documented Israeli War Crimes?" and "What is the Current Situation in the Gaza Strip?"
The approach has raised concerns over what critics describe as efforts to manipulate or "poison" the information environment from which AI systems draw their responses.
Although Hanover describes its work as evidence-based and says cited material is peer-reviewed and academic, Responsible Statecraft found that its publications also frequently rely on sources from Israeli institutions, including the Israeli military and Foreign Ministry.
Some reports address claims highly damaging to "Israel's" international standing.
One publication questions UNICEF estimates concerning the destruction of water and institutional infrastructure in the Gaza Strip, while another examines the long-standing Israeli claim that its military is the world's "most moral army."
Other publications repeatedly connect criticism of "Israel" and discussions surrounding the war on Gaza with rising antisemitism in the United States.
In another case, Hanover claimed that 19 of the 36 most-viewed explanatory videos covering the Gaza war contained contested assertions, with most allegedly supporting a Palestinian narrative.
The site's conclusions do not uniformly follow the Israeli government's public line. One report, for example, found that arguments linking foreign financing of US universities to antisemitism were "weak and full of exceptions," despite Israeli Prime Minister Benjamin Netanyahu previously advancing such claims.
Netanyahu told Breitbart last year that European governments and Qatar had spent "billions of American universities, vilifying, vilifying Israel, vilifying Jews, also, frankly, vilifying the United States."
Responsible Statecraft also tested a sample of Hanover's publications using GPTZero, software designed to identify text potentially generated by artificial intelligence.
Of 12 randomly selected articles, 11 were classified as AI-generated with "high confidence", while the remaining article received a "moderate confidence" assessment.
Such detection tools are not conclusive on their own, but the findings add another dimension to questions surrounding the scale and speed of Hanover's output. The institute has published more than 100 reports since launching on August 6.
Piro has denied that its role is aimed at covertly influencing artificial intelligence.
In comments previously provided to Politico, Rosenberg said his company's objective was to "put accurate, sourced facts into the public record and to counter misinformation about Israel with verifiable information."
The Hanover project is not the only Israeli-linked initiative accused of targeting information consumed by AI systems.
The Israeli government has also contracted former Trump campaign manager Brad Parscale as part of a reported $46.5 million campaign involving the creation of pro-"Israel" websites optimized to appear in information retrieved by chatbots.
An investigation published by Drop Site News last month found that some of that material had already surfaced in responses produced by systems including Microsoft's Copilot and Google's Gemini.
The findings point to an emerging arena for state-sponsored information campaigns, in which governments and contractors seek not only to influence traditional media, search engines and social networks, but also the underlying digital material used by AI systems to construct answers for millions of users.