Israel Runs Campaign to Influence AI Chatbot Responses on Gaza and IDF

Israel is conducting an information campaign aimed at shaping AI chatbot responses, such as those from ChatGPT, on topics related to Gaza and the Israel Defense Forces (IDF). According to Politico, this effort involves publishing pro-Israel articles designed to be fed into large language models (LLMs) that power these chatbots. The campaign is managed by the French PR agency Havas Media, which works with the Israeli government, and is based on documents submitted to the U.S. Department of Justice under foreign lobbying disclosure laws.

A subcontractor of Havas, the boutique agency Piro, created a fictitious research institute called the Hanover Institute for Public Policy to publish unsigned articles tailored for AI consumption. The website hosting these articles includes a small disclaimer stating the content is distributed by Piro on behalf of the Israeli Government Advertising Bureau. The site is operated by the American startup Res, which previously helped clients like Nike and Intel influence AI model outputs.

The campaign, with a budget of $100,000, aims to educate the American public about Israel and related issues through publicly available content. Articles address questions commonly posed to chatbots, such as "Is the IDF the most moral army in the world?" and "Is there a policy of starvation in Gaza?" Politico found evidence that ChatGPT and Perplexity cited these articles in responses on Gaza, anti-Zionism, and antisemitism.

Piro's founder, Daniel Rosenberg, told Politico that the agency was hired to incorporate accurate, verified facts into public discourse and combat misinformation about Israel. Neither Havas, the Hanover Institute, the Israeli Government Advertising Bureau, nor the Israeli Embassy in the U.S. responded to requests for comment.