Piro Files 560,000-Word Fake Thinktank Under FARA as Israel Pushes AI Influence
Updated
Updated · The Guardian · Aug 26
Piro Files 560,000-Word Fake Thinktank Under FARA as Israel Pushes AI Influence
3 articles · Updated · The Guardian · Aug 26
Summary
Piro Inc disclosed to the US Justice Department that it distributed material for Israel through the Hanover Institute for Public Policy, a thinktank-branded site with no evident legal entity, staff, address or bylines.
124 reports totaling more than 560,000 words appeared on the site in nine days, framed as neutral research but repeatedly softening allegations over Gaza, genocide and apartheid while posing chatbot-style questions in nearly every headline.
An llms.txt file initially matched Res, a platform that markets content designed to be cited by ChatGPT, Perplexity, Claude and Gemini, before being replaced; Piro separately advertises "AI Story Optimization" for how LLMs judge credibility.
The filing sits inside a wider Israeli-funded influence effort routed through Havas Media and other contractors: FARA records show $15 million to Clock Tower X, nearly $9.9 million to Targeted Communications Global and $1 million to Piro.
The campaign comes as US views of Israel worsen—Gallup found 36% sympathized more with Israel in February—and researchers warn the bigger risk is seeding AI training data that can shape chatbot answers without visible sourcing.
If a fake think tank can trick AI, how many other global narratives have already been rewritten by machines?
Will the rise of AI Story Optimization make it impossible to distinguish genuine research from state-funded propaganda?
Israel’s Multi-Million Dollar AI Influence Campaign: The 2026 Hanover Institute Data Poisoning Exposed
Overview
In August 2026, an investigation revealed that the Hanover Institute, posing as an independent think tank, was actually a front for a multi-million dollar Israeli government campaign to influence AI systems. Funded through Piro, Inc. and Havas Media, the Institute rapidly published over 100 reports designed to mimic credible research, which were then absorbed by major AI platforms like Copilot and Gemini without user warnings. This operation marked a shift from targeting people to targeting the algorithms that shape public knowledge, exposing major vulnerabilities in AI regulation and sparking international backlash and new European efforts to counter such manipulation.