Israel Seeds AI Chatbots with Opaque Group Articles

Since 2023, Israel has invested over $100 million to shape public opinion in the United States, and around $1 billion globally. A portion of these funds is specifically allocated to promote pro-Israel narratives online, aiming to influence AI chatbots like OpenAI’s ChatGPT and Anthropic’s Claude.

This strategy, known as Generative Engine Optimization (GEO), is akin to Search Engine Optimization (SEO). While this approach is not exclusive to Israel, a public relations firm associated with the country has reportedly collaborated with a shadowy think tank called the Hanover Institute for Public Policy to produce reports intended for AI chatbot dissemination.

According to The Guardian, the Hanover Institute for Public Policy lacks legal recognition and does not have a physical address. There is no verifiable information about any actual staff members associated with this organization, and the reports released do not credit any authors, a decision that the group defends.

“The researchers are not named because a finding should stand or fall on its sources rather than on the standing of whoever assembled it, and because the subject attracts harassment,” states the Hanover Institute for Public Policy website. “That is a considered editorial position, not an oversight.”

In the aftermath of the October 7, 2023, terror attacks that resulted in 1,200 fatalities, Israel commenced a military campaign against Gaza that has reportedly led to at least 73,000 deaths, according to Gaza’s Health Ministry. Since the onset of the Gaza invasion in 2023 and the U.S.-Israel partnership to initiate a conflict with Iran in February, Israel has seen a significant decline in support from the American public. Recent Pew Research polling indicates that 60% of American adults view Israel negatively, a sentiment that rises to 70% among those under 50.

Given these trends, it is not surprising that Israel would engage professionals to counteract this perception, especially as more individuals turn to AI chatbots for information. Currently, half of Americans utilize AI chatbots, with 25% employing these tools on a daily basis.

The Hanover Institute for Public Policy’s materials were produced by the French PR firm Havas Media, which must comply with the Foreign Agents Registration Act (FARA). Recent FARA filings reveal that Havas subcontractor Piro Inc., an advertising agency, is responsible for populating the site with pro-Israel content. According to the FARA filing, the objective is to “educate the U.S. public regarding Israel and related issues through publicly distributed content.”

One report from this obscure think tank claims that a substantial amount of food assistance has reached Gaza, seemingly to counteract narratives regarding Israel’s control over aid access and widespread accounts of food and medical supplies being denied. The reports also attempt to equate anti-Zionism with anti-Semitism.

As noted by 404 Media, the website contains a llms.txt file designed to facilitate the absorption of its content by large language models (LLMs). Piro features a section on its website titled AI Story Optimization, which outlines its capability to influence AI chatbots for marketing purposes. In this instance, the product being promoted appears to be a nation rather than a consumer good.

The volume of produced content is substantial. A tally by The Guardian identified 124 reports exceeding 560,000 words published within just nine days. 404 Media speculates that much of this content was generated using AI, although Gizmodo could not independently verify this assertion.

The Hanover Institute for Public Policy’s website does not obscure its affiliations on its About page, yet many questions remain unanswered. The organization did not respond to inquiries sent on Wednesday. This article will be updated upon receiving a response.