Looking at the allegations that Israel has been using AI to spread false information regarding the Gaza conflict and how this can be dangerous to democracy and our trust in AI. Israel has long tried to sway global opinion on its issues with Palestine. It has done this by hiring PR companies and using pro-Israel news sources and political figures to help their case. It appears Israel has gone one step further and is now trying to control what people will learn from A.I.

Nick Cleveland-Stout wrote an article for Responsible Statecraft on August 17th 2026, about the fabrication of a think tank known as the Hanover Institute for Public Policy. This fake think tank was made as part of a communication project for Israel.

Subsequent reporting by The Guardian exposed the operation's extraordinary scale. Between August 6 and 14, the Hanover Institute published 124 research-style reports containing more than 560,000 words. The documents addressed controversial subjects ranging from Israeli military operations in Gaza to Palestinian displacement, allegations of genocide, humanitarian conditions, and the political influence of the American Israel Public Affairs Committee (AIPAC).

It appeared their goal was not only to convince humans, but also to influence the answers of AIs such as ChatGPT, Gemini, Claude, and Perplexity.

This development raises a fundamental question: What happens when governments attempt to manufacture the information that artificial intelligence presents as objective knowledge?

The Creation of an Artificial Think Tank

It had all of the trappings of a real US think tank. Good graphics on their website.  Reports and papers with plenty of footnotes and well-thought-out policy ideas.

But it had some glaring omissions for a typical think tank. No researchers, no street address, and no bylines.

According to Cleveland-Stout's investigation, the website carried a disclosure indicating that Piro Inc., an American communications company, created it on behalf of the Israeli Government Advertising Agency, commonly known as LaPam.

The organization was therefore not an independent research institution in the conventional sense. It was a communications vehicle associated with a foreign government's public diplomacy campaign.

The scale of publication was particularly remarkable. Producing more than half a million words of supposedly authoritative research in nine days would be an extraordinary accomplishment for even a well-established academic institution.

Responsible Statecraft reported that an analysis of 12 randomly selected Hanover reports using GPTZero classified all 12 as likely AI-generated. Such detection tools are not conclusive proof of authorship, but the findings reinforced questions about how the material had been produced.

The more important issue, however, was not whether artificial intelligence helped write the reports. It was whether those reports were intended to influence other artificial intelligence systems.

Manipulating the Machinery of Knowledge

Modern AI chatbots generate answers by drawing upon patterns learned during training and, in some cases, information retrieved from the internet.

This creates a new opportunity for political influence.

Traditionally, propaganda required a newspaper, television network, social media platform, or influential commentator to distribute a particular narrative. Today, an organization can publish large quantities of apparently authoritative material in the hope that AI systems will retrieve, cite, or eventually absorb it.

Piro Inc. publicly advertised a service called "AI Story Optimization," describing its ability to produce content engineered around how large language models assess credibility.

The Hanover Institute's publications appeared particularly suited to this strategy.

The titles they created resembled questions you would ask an AI chatbot. Some included whether Israel was committing genocide in Gaza, whether Palestinians were being starved on purpose, and whether being anti-Zionist is considered Jewish hate.

The reports often used academic-sounding language while presenting interpretations that tended to downplay criticism of Israeli policies.

According to The Guardian, one 2022 study examining antisemitic Facebook comments was cited 454 times across 88 Hanover reports, including discussions only indirectly related to antisemitism.

It appears they are trying to make it seem as if pointing out Israeli actions will always have some connection to being anti-Semitic, no matter what it's being brought up about.

While these probes show an attempt to change how AIs answer questions, we can't say for certain they have manipulated all the major chatbot models.

Following the Money

The financial arrangements surrounding the project reveal a much larger international communications network.

According to American Foreign Agents Registration Act filings examined by journalists, Piro Inc. worked through Havas Media, an international advertising group involved in Israeli government communications.

The Guardian reported that Piro had contracts totaling approximately $1 million across two work orders.

Other American contractors associated with the broader campaign included Clock Tower X, a company led by Brad Parscale, a former senior campaign adviser to Donald Trump.

Reports said Havas spent around $15 million on Clock Tower X from October 2025 to March 2026. Another company, Targeted Communications Global, was paid about $9.9 million at the end of 2025.

While these payments could be attributed to other forms of Public Diplomacy and digital communications, they still relate to efforts to portray Israel in a favorable light around the world.

The Israeli government has defended its overseas communications activities. Piro co-founder Daniel Rosenberg has described his company's purpose as providing accurate, verifiable information to counter misinformation about Israel.

There is nothing wrong with a government trying to advocate for their stance to other countries. Especially when they pay someone to promote an idea that seems to come from an unbiased researcher. And don't properly disclose where the funding is coming from.

Gaza and the Struggle Over Historical Truth

The controversy becomes especially consequential because of the humanitarian catastrophe in Gaza.

The war has generated extensive documentation from United Nations agencies, international humanitarian organizations, journalists, and human rights investigators. These sources have examined civilian casualties, destruction of infrastructure, restrictions on humanitarian assistance, displacement, and alleged violations of international law.

They often made caveats about contested information.  Casualty counts are questionable, and some cases never reached a final verdict.

There are valid caveats when discussing international law. There should be a difference between claims.  Tentative decisions and verdicts. 

But repeating caveats that let certain criticisms be dismissed can look academic while leading you to a predetermined conclusion.

This is particularly dangerous when AI systems reproduce arguments without revealing the institutional interests behind them.

The distinction between antisemitism and criticism of Israeli government policy must also remain clear. Antisemitism is a serious form of prejudice that deserves unequivocal condemnation. But criticism of a government's military operations, occupation policies, or treatment of civilians is not automatically antisemitic.

This can blur the line between political discourse and anti-religious hate. Many more nations could be at risk from this.

While this article may be about Israel, it can affect many other areas of the world.

Countries such as China, Russia, India, Pakistan, and the US all have strong motivations to have AIs represent their history and conflicts.

Corporations could similarly attempt to shape chatbot responses concerning environmental damage, product safety, financial misconduct, or labor practices.

The danger is the industrialization of manufactured credibility.

A fake organization can produce hundreds of articles that point to real institutions. Mimicking proper research techniques and pushing certain results is cheap.

AI may be unable to detect this and give false information to users who want unbiased data.

This is not merely misinformation in its traditional form. It attempts to influence the infrastructure through which future information is discovered and interpreted.

Can Artificial Intelligence Defend Its Independence?

The investigations also offer grounds for cautious optimism.

The Guardian reported that when ChatGPT was asked about Hanover Institute publications, it provided links to reports while also warning users about controversies surrounding the organization's funding and legitimacy.

This shows AI can be smart enough to flag certain bad sources.

But as mentioned, these measures are not foolproof.

We need better source control from our AI. They need to detect astroturfing and distinguish a true research organization from an AI-funded political organization.

The government should also disclose any astroturfing from foreign countries. Researchers and journalists need to dig deeper into political money and its effect on the new wave of AI.

Users, meanwhile, should recognize that a chatbot's confident response is not necessarily an independently verified conclusion.

Conclusion: Propaganda in the Age of Intelligent Machines

The Hanover Institute affair represents an important warning about the future of information warfare.

The central concern is not that Israel seeks to defend its policies. Every government has the right to explain its position, challenge accusations, and participate in international debate.

The concern is that political advocacy may be deliberately disguised as independent scholarship and distributed in forms designed to influence artificial intelligence systems.

Another political irony comes into play here.  Polls from the Guardian show that Americans have grown to dislike Israel more than ever since the start of the Gaza conflict. Even after spending so much money on public relations.

This suggests that sophisticated communications campaigns cannot necessarily overcome public concern about the underlying policies themselves.

A government may influence what a chatbot says, but that does not mean it can change what people witness, experience, or believe.

The Hanover Institute controversy should therefore be understood as more than an Israeli public relations scandal. It is an early warning of a global struggle over who controls the information environment of artificial intelligence.

If manufactured research can be transformed into apparently authoritative AI-generated knowledge, the boundary between information and propaganda becomes dangerously blurred.

Preserving that boundary will be one of the defining challenges of the digital age.