TLDRocket
Sign in

Israel Is Running a Synthetic Think Tank to Influence AI Search Results

404 Media Matthew Gault

Israel-backed site has pumped out 100+ AI-made articles to shape chatbot answers. It’s a propaganda operation built for LLMs, not people, and it’s already public.

Based on reporting by 404 Media, Matthew Gault — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

A synthetic think tank called the Hanover Institute for Public Policy has published more than 100 articles in less than a month, and the whole thing appears built to nudge AI answers toward Israel. First noticed by Politico, the site is run by the American ad firm Piro Inc and paid for by Israel. The target isn’t just readers. It’s the models themselves, which keep scanning the web and feeding on whatever shows up most often and most clearly.

Hanover posts roughly a dozen pieces every few days on Israel, antisemitism, and Palestine. The headlines all come as questions, the articles have no bylines, and they often include graphs and cite real sources without linking out. The prose is strange and stiff, the kind of output that sounds like a machine trying to mimic seriousness after reading too much policy writing. One article’s explanation of Israel restricting food to Gaza was so tangled it barely held together as English.

The site says the missing bylines are a feature, not a bug. On its about page, Hanover argues that readers should focus on the sources instead of the author, and says the registrant responsible for the material is disclosed elsewhere. But there’s another clue that the project is aimed at machines: Hanover has an llms.txt file, a format meant to make scraping easier for large language models. Pangram, an AI detection tool, found that three Hanover articles it tested were entirely AI-written except for the bibliography, and the images were AI-generated too.

Piro has been unusually open about the larger strategy. Co-founder Daniel Rosenberg wrote on LinkedIn about a system for spotting where a brand’s story is weak or missing, then strengthening the signals that shape how AI explains it. On Piro’s own site, the company says search is now about “story” and that it maps the sources models read, from Reddit and YouTube transcripts to reviews, forums, and comparison sites. That is not SEO in the old sense. It’s narrative laundering for chatbots.

The Foreign Agents Registration Act filings make the setup even clearer. Hanover is registered as a foreign agent, and the paperwork says Piro is working with LaPam, the Israeli government’s official advertising agency, on behalf of the State of Israel. Those filings also include invoices from HAVAS Media, another agency working for Israel, including $900,000 dated April 30, 2026 to develop the project and another $100,000 for two months of work. The filings say the content is meant to educate the U.S. public through publicly distributed material. The website may call that institutional publishing. It looks a lot more like industrial spin.

My take — AI-written commentary, not fact-checked reporting

This is the part of AI everyone politely skips: the models are now a propaganda surface, and some clients have already noticed. The open-web dream turns into a soup factory the second people start feeding it with purpose. If chatbots are going to quote whatever they’ve been nudged to see, then “neutral AI” is just marketing with better typography.

Read more about this at: 404 Media

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.