Israel is reportedly using a US-based communications campaign to influence the online information that artificial intelligence chatbots may draw on while answering questions about Gaza, Palestine and the Israeli military.
The initiative is linked to the Hanover Institute for Public Policy, a website that describes itself as an American research body and has recently published a large volume of reports on the Gaza war, Palestinian displacement, antisemitism and Israel-related issues.
According to Politico, US Foreign Agents Registration Act filings show that Piro Inc., an American advertising firm working with Havas Media, a French communications company, was involved in a $100,000 project to produce and distribute material on Israel.
Rather than attempting to directly change the workings of AI platforms, the campaign appears aimed at increasing the availability of Israel-focused content online. Such material can be indexed by search engines and may later be cited by generative AI systems while responding to user queries.
Tests carried out by Politico reportedly found that ChatGPT and Perplexity referred to Hanover Institute reports in responses to questions involving Gaza, anti-Zionism and antisemitism.
An investigation by Responsible Statecraft said the Hanover Institute had uploaded more than 100 reports since August 6. The documents, which contain citations, statistics, footnotes and detailed formatting, do not identify individual authors.
The institute’s website discloses that it was created by Piro on behalf of Israel’s Government Advertising Agency, LaPam.
Responsible Statecraft reported that Piro had received $900,000 from the Israeli government for its work. The company also offers a service called “AI Story Optimization”, which it describes as developing content to suit the way large language models assess and surface information.
NewsGuard analyst Alice Lee told the publication that the reports were structured in a manner likely to appeal to people using online search and AI tools to understand the conflict, particularly because they contain extensive data and references.
The outlet used GPTZero to examine 12 Hanover Institute reports, finding that 11 were assessed as likely AI-generated with high confidence and one with moderate confidence. However, AI-detection tools cannot definitively establish whether a text was produced using artificial intelligence.
The reported project forms part of a larger Israeli outreach effort in the United States. The Wall Street Journal reported in July that Israel had signed a contract worth more than $45 million with Brad Parscale, a former adviser and longtime ally of US President Donald Trump.
Hanover Institute website showing research reports on Israel and Palestine. Photo: Screengrab
The wider campaign reportedly involves digital messaging, media partnerships, AI-generated content and other communication efforts aimed at improving Israel’s standing among US audiences.
Piro’s registration documents do not specifically state that its work for Israel is intended to shape AI responses.
Piro co-founder Daniel Rosenberg told Politico that the firm’s task was to place accurate, properly sourced information in the public domain and counter misinformation about Israel.