- Filed
- September 9, 2026, 22:00 UTC
- Actor
- Israeli Government Advertising Agency (LaPam) via contractors Piro Inc and Havas Media (FARA-registered foreign influence activity; state-directed public diplomacy)
- Target
- U.S. and English-language AI chatbot users, search-engine results, and broader public information environment on Israel-Palestine issues
- Confidence
- documented Primary documents, court records, official disclosures
- Techniques
- Synthetic mediaNarrative launderingCard stackingManufactured consensusAstroturfing
- Channels
- fake think-tank website
- AI-generated academic-style reports
- question-formatted content for LLM optimization
- FARA-disclosed PR subcontracting
- search-engine and chatbot citation pathways
Summary
Investigations by Politico, 404 Media, Responsible Statecraft, The Guardian and others in August–September 2026 document that the Hanover Institute for Public Policy is not a real think tank but a website created by New York firm Piro Inc under a roughly $900,000 contract routed through Havas Media for Israel’s Government Advertising Agency (LaPam). In nine days it published 124 AI-generated reports totaling over 560,000 words, structured as question-answer content optimized for LLM ingestion. Tests showed ChatGPT and Perplexity citing the material. FARA filings confirm the foreign-agent relationship.
Analysis
Primary documentation rests on FARA filings by Piro Inc and contemporaneous investigative reporting from Politico, 404 Media (25 August 2026), Responsible Statecraft (16–17 August), The Guardian (26 August) and subsequent coverage through early September. The Hanover Institute presents itself as an American public-policy research body yet has no physical office, named scholars, legal entity status or bylined authors. A website disclaimer and FARA paperwork explicitly link it to LaPam, Israel’s state advertising agency, via Havas Media Germany and Piro. In a nine-day burst beginning 6 August 2026 the site published 124 reports averaging roughly 4,500 words each, many framed as neutral answers to questions that users or retrieval systems are likely to pose (e.g., starvation policy in Gaza, moral status of the IDF, AIPAC funding). AI-detection tools flagged the bulk of sampled text as machine-generated; the volume and cadence are consistent with automated production optimized for indexing.
The mechanism is narrative-laundering through synthetic media at industrial scale. Authentic institutional markers—footnotes, charts, academic tone, selective citation of real primary sources—are layered onto content whose generation and dissemination are directed by a foreign government contractor chain. Card-stacking appears in the systematic framing that privileges Israeli government and IDF sources while omitting or minimizing contradictory international findings. The explicit commercial offering of “AI Story Optimization” by the contractor reveals intent to seed the training and retrieval corpora of large language models so that chatbot answers reproduce the desired framing without the user encountering an overt propaganda label. Astroturf and manufactured-consensus effects arise because the material appears as independent American research rather than state messaging.
This qualifies as an influence operation because authenticity is artificially constructed (fictitious institute, AI authorship, borrowed academic form), origin is partially obscured behind commercial intermediaries even while FARA disclosure exists, and the design goal is to distort the downstream information environment consumed by AI systems and their users. Technique families are scored independently of the actor’s identity; identical methods of synthetic content farms, institutional camouflage and LLM targeting have been observed across commercial, party and state actors. The operation’s visibility in FARA records and rapid researcher detection limit its stealth, yet the precedent—that governments can attempt to poison the knowledge base of widely used AI tools—is load-bearing for information integrity.
Literacy counter: treat any newly appearing “institute” or “research center” that lacks verifiable staff, physical presence, peer-reviewed track record or transparent funding as provisional, especially when its output clusters tightly around a single geopolitical narrative and is optimized for question-answer formats. Cross-check chatbot citations against primary documents, multi-source journalism and established academic or official records. Prefer original reporting, court filings, UN or NGO primary data and contemporaneous footage over any single branded report that cannot be independently attributed to a real institution. AI developers and search providers should surface provenance, foreign-agent status and coordination signals for synthetic institutional content rather than allowing it to enter retrieval corpora unexamined.
Evidence
- 404 Media: Israel Is Running a Synthetic Think Tank to Influence AI Search Results (25 Aug 2026)
- The Guardian: Fake US thinktank set up and funded by Israel sought to game AI for propaganda (26 Aug 2026)
- Responsible Statecraft: Israel creates fake think tank in likely attempt to dupe AI chatbots (17 Aug 2026)
- Al Jazeera opinion synthesizing FARA trail and chatbot citation tests (4 Sep 2026)
- Wikipedia summary of Hanover Institute for Public Policy (updated Sep 2026)