By Dr. İdris Kardaş
A few months ago, I wrote a piece in this column titled “Turkey in the Memory of Algorithms.” In it, I examined how artificial intelligence learns about the world, the sources it draws from, and—specifically—the perspectives from which it generates information about countries.
The central issue I highlighted in that article was this:
The way billions of people around the world access information is changing, and consequently, the methods countries use to present themselves to the world are undergoing a radical transformation. Today, what artificial intelligence systems know about a country is just as important as the information held in people’s minds—indeed, it is significant enough to actually reshape the knowledge within the human mind itself.
For instance, I had said that we urgently need to consider the sources from which people anywhere in the world learn about topics such as Turkey’s Cyprus policy or its activities in Africa, the genocide committed by Israel in Gaza, or any historical issue dating back further.
This is because people no longer tend to seek answers to these questions by browsing through dozens of pages on search engines; instead, they ask ChatGPT, Gemini, or Claude.
At the time, I had specifically drawn attention to Israel’s activities in this arena.
Documents and research that have emerged in the intervening months demonstrate that the issue is far more extensive than we had imagined. It is being revealed in concrete and clear terms.
Investigations published by The Guardian, Axios, and Drop Site News have detailed an organization established by Israel to influence the responses generated by artificial intelligence systems. The picture that emerges involves multi-million-dollar budgets, technology companies, and an extensive network of fabricated content disguised as academic research.
Let’s take a closer look…
In August, The Guardian published a fascinating investigation. According to the report, a website calling itself the “Hanover Institute for Public Policy”—and presenting itself as an American think tank—had published 124 research reports in just nine days. That amounts to over 560,000 words of content. Yet, virtually no information could be found regarding the organization’s researchers, physical address, or corporate history.
In contrast, the reports were prepared in a highly professional manner. They employed academic language and cited UN documents, human rights organizations, and various studies.
The titles of the reports and studies were exactly what you might expect: Is Israel committing genocide in Gaza? Were Palestinians driven from their lands? Did Israel commit war crimes?
As you can see, these are the kinds of questions millions of people might ask artificial intelligence. This is precisely where the logic behind the operation becomes apparent. First, the questions people are likely to pose to AI are identified. Then, content designed to answer these questions is carefully placed online. The researcher is fake, the sources cited in the study are fake, the data is fake, the witnesses are fake—everything is fake. The reason they are able to publish such a vast amount of text is that they have AI bots write it for them.
An investigation by The Guardian revealed that the Hanover Institute is linked to an American media company called Piro. Upon examining documents Piro submitted to the U.S. Department of Justice, they discovered that the company is a communications agency working on behalf of the Israeli government.
In carrying out this initiative, Israel analyzed which websites artificial intelligence systems selected as reliable sources. Essentially, they first examined which websites systems like ChatGPT, Gemini, or Claude deemed trustworthy, and then tailored their content accordingly. The reasoning was that as fake reports cited these specific websites, the AI would come to accept the information as credible.
In the marketing world, this is known as Generative Engine Optimization, or GEO.
I have previously written articles touching upon this concept. In the Google era, websites achieved high rankings in search results through SEO methods. GEO, however, aims to influence which sources and websites are utilized in answers generated by artificial intelligence.
The distinction is significant. For instance, in the past, you could choose from the various links that appeared on Google. Artificial intelligence, on the other hand, presents information compiled from diverse sources as a single, unified answer. Often, the user is unaware of how the underlying sources for that answer were selected, yet they accept the information—which arrives without a specific attribution—as accurate.
Another notable name of the organization is Donald Trump’s former campaign manager, Brad Parscale. Clock Tower X, the company he founded, plays an important role in digital communication activities carried out on behalf of Israel.
According to records at the US Department of Justice, the company works on behalf of Israel. According to Axios’ research, Parscale’s team created nine different websites so that artificial intelligence systems would produce more positive answers about Israel.
One of the sites describes Israel as a country that strives for peace and coexistence. Another highlights military cooperation between the United States and Israel. Others address the legal aspects of operations in Gaza or Hamas’s international status. Of course, always in favor of Israel.
Each site seems to specialize in a particular topic. Thus, artificial intelligence sees this: Content coming from different sources but prepared within the framework of the same communication strategy.
Now let’s come to the most critical point. Do Israel’s efforts work? So, when someone anywhere in the world asks what happened in Gaza, is there an artificial intelligence system that gives answers in favor of Israel?
The research published by Drop Site News in July revealed important findings on this subject. Accordingly, most of the websites created by Clock Tower X in favor of Israel were included in the Internet archive called Common Crawl.
Common Crawl is one of the key data sources used in the development of large language models. However, inclusion in this archive does not, of course, mean that the content in question is used to train every single AI model. Nevertheless, significant developments have been observed.
For instance, when researchers asked Perplexity about the benefits of military cooperation between the US and Israel, they found that a website called Allyvia—created by Clock Tower X—was cited as one of the primary sources. Microsoft Copilot was also observed referencing content from the same network. This is truly critical; a single drop of pollution introduced into the ocean of information was enough to contaminate the entire ocean.
Let us turn to the economic aspect of this matter.
According to research published by the Quincy Institute, the Israeli government has spent over $1 billion on public diplomacy activities worldwide since October 7, 2023. Of course, not all of this money was allocated to artificial intelligence initiatives; the figure encompasses social media campaigns, advertisements, lobbying efforts, and other communication activities.
However, activities aimed at influencing the responses of artificial intelligence systems have become a significant part of this operation. The term used by the Israeli bureaucracy to describe these efforts is also noteworthy: cognitive warfare.
Information presented to the parliament by Israel’s Ministry of Diaspora Affairs mentions the creation of dozens of websites and the monitoring of how artificial intelligence models react to this content. In other words, Israel is not doing this covertly.
Let us conclude the article by looking at the context of Turkey.
In the piece titled “Turkey in the Memory of Algorithms,” I asked the following question:
How do ChatGPT, Claude, or Gemini perceive Turkey? From which sources do they learn about Turkey? Whose narrative shapes their understanding?
The documents that have come to light today better illustrate why these questions are significant.
Imagine a university student anywhere in the world researching Turkey. In all likelihood, instead of meticulously examining Turkey’s official statements, the work of Turkish academics, and historical documents, they will simply ask an artificial intelligence a question.
So, to what extent will studies originating from Turkey feature in the answers generated by the system? To what degree will academic texts published in Turkish be utilized? How will Turkey’s perspective be represented in research conducted in English, French, or German? How will we be able to counter the information disseminated by the millions of people spreading disinformation about Turkey?
The same questions apply to Turkey’s Africa policy, its activities in Syria, its defense industry, and historical issues.
I had previously touched upon the importance of perspective—drawing on the work of Edward Branigan—and used the phrase, “The owner of the story is also the owner of the truth.”
In the age of artificial intelligence, this statement takes on a new meaning, as algorithms are increasingly playing a role in how stories are told.
Israel’s initiatives illustrate how states are beginning to utilize this new arena. After all, in the future, people will increasingly come to understand countries, societies, and historical events through answers generated by artificial intelligence.
The sources behind those answers will shape the world’s collective memory, making it crucial for us to prepare for this reality today.
Source: haberturk.com
