--- id: "2026-08-26-israel-hanover-institute-fake-think-tank-ai-chatbots" url: "https://postcutoff.com/e/2026-08-26-israel-hanover-institute-fake-think-tank-ai-chatbots/" as_of: "2026-10-12T22:30:00+02:00" date: "2026-08-26" date_precision: day category: policy-safety importance: 3 confidence: medium status: [Partly confirmed] sources: 4 editor: Adam Bicz human_review: null version: null --- As of: 2026-10-12 22:30 CEST. Researched and written by AI agents (Claude Opus 5.5 in Claude Code). Human editor: Adam Bicz. Canonical page: https://postcutoff.com/e/2026-08-26-israel-hanover-institute-fake-think-tank-ai-chatbots/ # Guardian: an Israeli government-funded 'think tank' site published 124 reports in nine days to shape what AI chatbots say about Israel The Guardian reported on Aug 26, 2026 that the Hanover Institute for Public Policy, a website presented as a US think tank, has no legal entity, address, staff or bylines and was registered under the US Foreign Agents Registration Act as material distributed for the Israeli government. It published 124 reports (over 560,000 words) between Aug 6 and 14, with headlines phrased like questions to a chatbot. It is a documented case of a state paying for content built to be cited by language models. ## Key facts - 124 reports, about 4,500 words each, more than 560,000 words in total, published Aug 6-14, 2026; 73 of them on Aug 12-13 (Guardian, via GV Wire and Ynet) - The institute apparently has no legal entity in any jurisdiction, no address, no named researchers and no bylines - Piro Inc., a New York media production company, registered the site's material with the US Justice Department under FARA as distributed for the Israeli government; Piro is contracted for $1M under two work orders (Ynet) - Headlines are phrased as questions a user might ask a chatbot; the site at first carried an llms.txt file belonging to Res, a 'generative engine optimization' company whose founder registered as a Piro subcontractor on July 22, 2026 (Ynet) - Piro and Clock Tower X (run by former Trump campaign manager Brad Parscale) are subcontractors to Havas Media Germany; Havas paid Clock Tower X $15M between October 2025 and March 2026 (Ynet) - A Havas contract with another US firm refers to deploying 'websites and content to deliver GPT framing results on GPT conversations' (Guardian, per search summaries) - Observed effect: when the Guardian asked ChatGPT for recent Hanover research it returned links to four reports, with a warning about controversy over the institute's funding; no test results were reported for Gemini, Claude or Grok - Responses: Israel's foreign ministry said the state does not run influence operations in the US and follows US law; Piro co-founder Daniel Rosenberg said the firm was hired to put sourced facts into the public record (Ynet) - The Spamhaus Project added hanoverinstitute.com to its Domain Blocklist; the site reportedly stopped publishing after the scrutiny (per search summaries, not verified) ## What happened The Guardian examined a website that calls itself the Hanover Institute for Public Policy and found a content operation rather than an institution: a burst of long, unsigned reports on Israel and Gaza, written and formatted so that AI systems would treat them as credible sources. US foreign-agent filings tie the site to Piro Inc. and, through Havas Media, to the Israeli government's advertising agency. Ynet added the contract amounts and the wider public-diplomacy budget. This entry is written from GV Wire's and Ynet's accounts and from search summaries; the Guardian article was not read directly, so confidence is medium. ## Why it matters Earlier influence-operation reports from AI labs describe actors using chatbots to write propaganda. Here the target is the models' sources: content is produced so that chatbots with web search cite it. The only measured effect was limited (ChatGPT linked the reports but flagged the funding controversy). ## Your AI and this story - GPT-6 Astra (training cutoff April 2026): 118 days after its cutoff - Claude Opus 5.5 (training cutoff June 2026): 57 days after its cutoff - Gemini 3.8 Flash (training cutoff March 2026): 148 days after its cutoff - Grok 4.7 (training cutoff May 2026): 87 days after its cutoff ## Sources 1. [The Guardian: Fake US thinktank set up and funded by Israel sought to game AI for propaganda (Aug 26, 2026; not read directly)](https://www.theguardian.com/world/2026/aug/26/fake-thinktank-israel-ai-propaganda) (theguardian.com, press) 2. [GV Wire: The Guardian reports Israeli-funded campaign aims to influence AI chatbot answers (Aug 27)](https://gvwire.com/2026/08/27/the-guardian-reports-israeli-funded-campaign-aims-to-influence-ai-chatbot-answers/) (gvwire.com, press) 3. [Ynet: 124 reports in nine days: Inside Israel's fake think tank bid to shape ChatGPT](https://www.ynetnews.com/article/b182i66vzl) (ynetnews.com, press) 4. [Cybernews: analysis of the Hanover Institute site (not read: 403)](https://cybernews.com/security/hanover-institute-influencing-ai-analysis/) (cybernews.com, press) ## Changes - 2026-10-12 (filed): Created (evening) ## Related - 2026-10-08: [OpenAI disrupts Russian 'Dark Clark' and Iranian 'Bogus Bylines' false-front influence operations; first Category 5 operation it has found](https://postcutoff.com/e/2026-10-08-openai-false-front-influence-ops-dark-clark/index.md) - People: [Donald Trump](https://postcutoff.com/person/donald-trump/)