Israel has reportedly expanded its public-relations campaign into the emerging world of artificial intelligence, as support for the country declines among Americans. According to reporting cited by RT, material connected to an Israeli-funded influence operation has appeared as a source in responses from ChatGPT and Perplexity, raising questions about how governments and political campaigns may attempt to shape what AI systems tell their users.
INR Report: Based on reporting by RT HERE, citing an investigation by Politico.
A New Front in the Information War
Israel has reportedly spent tens of millions of dollars on campaigns intended to improve its image in the United States, with Politico now detailing an initiative that could potentially influence the information encountered by artificial-intelligence systems.
According to RT's account of the Politico investigation, one $100,000 campaign involves the Hanover Institute for Public Policy, a website publishing reports dealing with antisemitism, Gaza, Israel and geopolitical issues.
The project was reportedly established through French public-relations company Havas Media and subcontractor Piro Inc on behalf of the Israel Government Advertising Agency, known as LaPam.
Havas Media's US influence work for Israel is registered under America's Foreign Agents Registration Act, or FARA.
Content Designed for the AI Age?
The Hanover Institute describes its work as documentary and says its research is academic.
Politico, however, reportedly identified FARA documents filed with the US Department of Justice connecting more than a dozen Hanover articles with a public-relations campaign producing what the filings describe as factual, source-supported informational material intended to educate Americans about Israel.
Of particular interest is the way some of the website's material is presented.
Reports carry question-based titles including subjects such as whether torture of Palestinian prisoners constitutes a war crime and who should be regarded as the aggressor in the Israeli-Palestinian conflict.
Politico suggested that this format could make the material more likely to be discovered and incorporated into responses generated by large-language models.
The investigation also reportedly identified technical markers connecting the Hanover website with Seattle AI company Res, which helps organisations improve the likelihood that they will be recommended or surfaced by AI models. According to RT, Res CEO Hai Tran confirmed the connection to Politico.
ChatGPT and Perplexity Cited the Material
There is an important distinction in the evidence.
According to RT, Piro's FARA filings do not explicitly state that influencing large-language models is an objective of the campaign.
Politico nevertheless reported that when it submitted neutral questions concerning Gaza and Israel, both ChatGPT and Perplexity returned answers citing material published by the Hanover Institute.
That does not establish that either AI company was involved in, endorsed, or knowingly facilitated the Israeli campaign.
Rather, it demonstrates a broader challenge facing generative AI: models and AI search systems can encounter publicly available information created as part of political, commercial or government influence campaigns.
If such material appears authoritative and is structured effectively for online discovery, it may potentially become part of the information ecosystem from which AI-generated answers are constructed.
Israel's Wider Influence Campaign
RT reports that this is not Israel's first Havas-linked campaign involving artificial intelligence.
Citing a Wall Street Journal investigation, RT says a previous $46.5 million campaign involved Brad Parscale, the former campaign manager for US President Donald Trump. That initiative reportedly included AI-generated pro-Israel messages aimed at Americans, alongside websites and posts intended to encourage more favourable responses from AI platforms.
RT also cites previous reporting that Havas Media managed a $900,000 influencer campaign producing Israel-friendly content.
Taken together, the campaigns illustrate how political communications are moving beyond conventional advertising, television and social media into an environment where organisations increasingly seek visibility within AI-generated information.
American Support for Israel Declining
The campaign comes amid a significant shift in American public opinion.
RT cites a June Pew Research Center poll reporting that six in ten US adults viewed Israel somewhat or very unfavourably, compared with 53% the previous year and 42% in 2022.
Israeli Prime Minister Benjamin Netanyahu reportedly recorded even weaker numbers, with 65% expressing little or no confidence in his policies.
Political opposition has also increased.
More than 100 Democratic members of the US House reportedly supported an amendment last month that would have cut $3.3 billion in American assistance to Israel. Although unsuccessful, RT notes that only 37 House Democrats had supported a comparable effort little more than two years earlier.
The Bigger Question: Who Shapes AI?
The significance of the Politico investigation extends well beyond Israel and Gaza.
Governments, corporations, political organisations and public-relations firms have spent decades attempting to influence search-engine results and social-media discussion. Artificial intelligence creates another target.
Instead of persuading only human readers, organisations now have an incentive to publish large quantities of carefully structured material that may also be discovered, indexed, cited or otherwise incorporated by AI systems.
This creates a new form of information competition: optimising political narratives not simply for people or search engines, but potentially for machines that millions of people increasingly rely upon for answers.
It also reinforces why AI-generated information should not automatically be treated as politically neutral simply because it is produced by a machine. The quality of an answer ultimately depends in part on the quality, provenance and diversity of the information available to the system.
Does This Affect New Zealand?
Yes — because New Zealanders increasingly use AI systems to understand international affairs, politics, public policy and breaking news.
If foreign governments and political organisations are deliberately creating material intended to become more visible to AI systems, the issue becomes one of media literacy as much as technology.
The appropriate response is not to assume that AI platforms are secretly controlled by governments, nor to assume that every cited source represents manipulation. The evidence presented here does not establish either proposition.
Instead, it highlights the importance of transparency about sources, distinguishing independently verified information from advocacy, and consulting competing perspectives when dealing with contested political subjects.
As AI becomes a gateway through which people understand the world, an old question of journalism is becoming equally important for artificial intelligence:
Who produced the information, who paid for it, and what were they trying to achieve?
Independent reporting. Original context. Credited sources.