Pro-Kremlin AI Disinformation Network Linked to Israel-Based Think Tank, Der Spiegel Reports
BERLIN — A sophisticated Russian anti-Western disinformation network that leveraged artificial intelligence to generate pro-Kremlin propaganda has been operationally linked to the International Burke Institute, an Israel-based think tank, according to a German news investigation published Tuesday. The report from Der Spiegel outlines how user accounts flagged and subsequently banned by OpenAI’s security teams were used to produce large volumes of Russian-aligned messaging, primarily for distribution across Telegram channels. The exposé alleges that the think tank—which describes itself as an “expert community” founded in 2025 and headquartered in Israel—sits at the center of a web of influence operations designed to undermine Western institutions, erode public trust in democratic governments, and amplify narratives favorable to Moscow. The investigation arrives at a time of heightened concern about the weaponization of generative AI tools by state-linked actors seeking to manipulate public opinion across Europe and North America. While the accounts operated under the guise of independent analytical voices, the content they produced consistently aligned with Kremlin talking points, casting NATO as a deceptive force and portraying the European Union as a doomed project. Neither the International Burke Institute nor a related charity named in the report, Dor Moriah, responded to requests for comment, Der Spiegel said.
According to Der Spiegel’s findings, the disinformation operation was uncovered through a digital trail connecting the accounts banned by OpenAI to the International Burke Institute’s promotional footprint. The institute, which projects the image of a policy-focused expert community, appears to be a front for a coordinated effort to lend analytical credibility to pro-Russian narratives across the German-speaking internet and beyond. Crucially, the news magazine reported that online evidence connects the institute to Dor Moriah, a charitable organization operated by Israelis of Russian and Ukrainian descent. Der Spiegel said that Dor Moriah has alleged ties to the Russian Foreign Ministry, raising the possibility that the operation extends from the highest levels of the Russian state into a network of ostensibly independent civil-society actors in Israel. The exact nature of the relationship among the think tank, the charity, and Russian officials remains unclear, but the linkage suggests a deliberate attempt to obscure the operation’s origins by laundering influence through a geographically diverse set of entities. The use of an Israel-based hub may also be intended to complicate attributions of Russian involvement, creating confusion among researchers and threat intelligence analysts. Der Spiegel’s reporting emphasizes that while the entities deny any wrongdoing—having thus far declined to answer questions—the pattern of behavior, content themes, and technical metadata consistently point toward a coordinated foreign-influence campaign.
OpenAI, the developer of ChatGPT, confirmed on Tuesday that it had taken decisive action against the accounts involved in the scheme. In a statement, the company said, “We recently banned a cluster of ChatGPT accounts originating in Russia that were being used to promote the International Burke Institute.” The statement further confirmed that the accounts were part of a covert influence operation spanning multiple languages, reinforcing Der Spiegel’s reporting that the activity was not a one-off mistake or simple macro-assisted spam but rather a deliberate, organized initiative. OpenAI’s security teams identified the accounts after detecting patterns of automated content generation and coordinated distribution that are consistent with known influence-operation tradecraft. The company did not disclose the full scale of the operation, but it did acknowledge that beyond boosting the International Burke Institute, the accounts were responsible for generating posts for several Telegram channels. This admission highlights the growing role of generative AI in contemporary disinformation campaigns, allowing small teams of operators to produce volumes of text that would normally require a large editorial staff. By automating the writing process, the operatives behind the network were able to maintain multiple Telegram channels with frequent, thematically consistent postings, creating the illusion of organic grassroots commentary. The fact that the AI-generated content was created in Russian and then repackaged for European audiences suggests a carefully staged process of message development and adaptation.
One of the Telegram channels identified in the investigation is called “Lahme Ente,” which translates as “Lame Duck.” The channel has approximately 20,000 subscribers and publishes a steady stream of content that is sharply critical of the European Union, the German government, and Western military support for Ukraine. According to Der Spiegel, recent headlines from the channel include provocative titles such as “What Is NATO Hiding From Us?” and “Europe’s Inevitable Downfall.” The channel’s material appears engineered to stoke public suspicion of established institutions and to amplify narratives that frame Western policy as imperialistic, incompetent, or actively hostile to the interests of ordinary citizens. The use of a German-language channel with a colloquial name suggests a targeted effort to influence German public opinion, which holds particular strategic significance given Germany’s role as a key NATO member and a major supplier of financial and military aid to Ukraine. Thematically, the content draws on familiar Kremlin narratives that have circulated in Western far-right and far-left populist spaces, including claims that NATO is expanding aggressively, that European leaders are hiding uncomfortable truths, and that the European project is destined for collapse. By injecting these narratives into a Telegram channel that appears independent, operators hope to bypass media gatekeepers and reach segments of the population that have grown distrustful of mainstream journalism. The AI-generated nature of these posts makes it even harder for platforms and researchers to detect the operation, as the text lacks the linguistic idiosyncrasies that often betray translation or non-native writing.
The investigation into the International Burke Institute and its associated Telegram channels is the latest in a series of public disclosures exposing Russian efforts to exploit social media and messaging platforms for influence purposes. Since the full-scale invasion of Ukraine in 2022, Western intelligence agencies and tech companies have repeatedly documented covert campaigns aimed at weakening public support for Kyiv, dividing NATO allies, and sowing discord within European societies. Historically, these operations relied on troll farms and shell media outlets staffed by operators in Russia or allied countries. However, the emergence of generative AI has lowered the barrier to entry for influence actors, allowing them to generate content at scale with minimal human oversight. OpenAI’s decision to publicly acknowledge the ban, and to name the International Burke Institute, represents an unusual step toward transparency, signaling that the company is willing to engage with researchers and journalists in exposing malicious activity on its platforms. Nevertheless, the incident raises unsettling questions about the broader ecosystem of AI-generated influence content that may remain undetected. Because ChatGPT and other large language models are available globally, malicious actors can access them legally and cheaply, generating text in dozens of languages with remarkable fluency. The result is an evolving threat landscape in which traditional attribution methods are often inadequate, and where the distinction between authentic and manufactured opinion becomes increasingly blurred.
The broader implications of this affair are profound for the credibility of think tanks, the integrity of public discourse, and the security environment in Europe and the Middle East. The International Burke Institute, by presenting itself as an expert community in Israel, had the capacity to lend a degree of analytical legitimacy to Russian-aligned narratives that would otherwise be dismissed as crude propaganda. If, as Der Spiegel suggests, the institute is linked to Dor Moriah and through that charity to the Russian Foreign Ministry, then the affair represents a deliberate effort to weaponize civil-society structures as instruments of foreign-policy messaging. For Israel, the connection poses a delicate diplomatic challenge, as the country must address the presence of Russian-linked influence activities operating from its territory without alienating the large Russian-speaking population that has emigrated to the country over the past three decades. For Europe, the affair reinforces the urgent need for coordinated media literacy programs, cross-border investigative mechanisms, and platform accountability standards. The response from OpenAI, though welcome, must be replicated across the technology industry, with companies moving beyond reactive bans to proactive identification of coordinated AI-driven influence networks. Meanwhile, Telegram’s role as a distribution platform continues to draw scrutiny, as the messaging app’s hands-off moderation policies have made it a favored haven for disinformation campaigns of all kinds. As the 2026 electoral cycle approaches in several European countries, the exposure of this network serves as a stark reminder that foreign adversaries are investing heavily in the manipulation of public opinion, and that the battle for the information domain is no longer fought solely by human agents but increasingly by algorithms designed to imitate human voices. The full extent of the damage caused by the International Burke Institute network may never be precisely measured, but its discovery underscores a sobering truth: in the age of artificial intelligence, any source—however authoritative it appears—may be a phantom engineered to deceive.