UNIT.City — місце, де люди працюють... КРАЩЕ! Обирай свій простір просто зараз 👉
Олександр КузьменкоWar
24 June 2026, 10:16
2026-06-24
Russian intelligence services are trying to poison artificial intelligence and Google data using fake Wikipedias
The Russian special services and the disinformation structures they control are shifting their focus from banal spam on social media to systemic «data poisoning.» They are trying to create an alternative information ecosystem of fake think tanks, media outlets, and Wikipedia clones. The main goal of this campaign, called «Project 2026,» is to force Google’s search algorithms and popular AI-based chatbots to present pro-Russian narratives to users as verified facts.
The Russian special services and the disinformation structures they control are shifting their focus from banal spam on social media to systemic «data poisoning.» They are trying to create an alternative information ecosystem of fake think tanks, media outlets, and Wikipedia clones. The main goal of this campaign, called «Project 2026,» is to force Google’s search algorithms and popular AI-based chatbots to present pro-Russian narratives to users as verified facts.
This is stated in internal documents of the Moscow agency «Social Design Agency» (SDA), which were obtained by Bloomberg .
SDA is a Russian company already under sanctions by the US, UK and EU for organizing influence operations and creating fake media (including the notorious Doppelgänger campaign). The new leak of 73 files contains detailed reports, chats and project proposals for the period from May 2023 to April 2026. They indicate that Moscow has moved from the chaotic «troll factories» of the Yevgeny Prigozhin era to methodological cyberattacks on the sources of knowledge that AI uses for training.
Cloning Wikipedia and optimizing for AI
Instead of simply posting on X (Twitter), Russian technologists create copies of Wikipedia («forks») and fill them with distorted information. For example, three cloned sites were created for Armenia before the elections (including spyurk.cyou, sevan.info, and khachkar.info), which copied articles from the online encyclopedia, but cut out references to the collapse of the USSR and added fabricated incriminating material about local politicians. To hide the Russian footprint, the infrastructure was registered allegedly through Turkish providers.
An even larger project was launched against Germany. According to the SDA’s plans of January 15, the Russians created about 200,000 websites of this nature. Their task is to edit at least 100 articles per month for search engine optimization (SEO) and «train» six large AI platforms with pro-Russian narratives every month.
«Their approach is to try to disable search engines by filling the space with content that cross-references their content or their narratives. This is an indirect way to infiltrate popular chatbots and search engines,» explains Kateryna Sedova, a non-resident senior fellow at the Atlantic Council’s Eurasian Center and a former employee of the US State Department.
The evolution of disinformation: from quotas to KPIs
The leaked documents demonstrate that Russia’s modern disinformation machine operates like a classic Western consulting firm, with clearly defined key performance indicators (KPIs), case studies, and sophisticated audience monitoring systems.
The Kremlin’s leader in these processes is Sofia Zakharova, an employee of the Information and Communications Department of the Russian Presidential Administration. In internal chats, she is called «Christine Keeler» (a reference to the British model whose spy scandal in the 1960s undermined the British government). Zakharova herself often refers in chats to coordinating budgets with «SVK» — these initials, according to the US Department of Justice, belong to Sergei Kiriyenko, the first deputy head of the Kremlin administration.
In addition to planning, the SDA carefully monitors the reach of its fakes. In particular, in September 2025, they analyzed the spread of a fabricated story that Ukrainian President Volodymyr Zelensky allegedly bought his mother luxury apartments in Dubai. According to their reports, this lie received 86 million views, of which 10 million were provided directly by 19 «contractors» of the Russian project (known in analytical circles as Storm-1516).
What experts and developers say
Jacob Rogers, deputy general counsel of the Wikimedia Foundation, expressed doubt that such clones could survive for long without the ongoing support of a community of real editors: «Forks usually don’t have long-term success. They can take a snapshot of Wikipedia at a certain point in time, but they can’t keep the information up-to-date.»
At the same time, cybersecurity experts note that English-language AI models have a higher degree of protection and better filters against disinformation, but models in less common languages (German, French, Armenian, or Ukrainian) are vulnerable to «data poisoning» methods due to the lack of high-quality local content.
Як нейромережі бачать вільну та незалежну Україну? Тест dev.ua
Нейронні мережі для генерації зображень бачать світ по-своєму, їхню логіку зрозуміти часом зовсім неможливо. Але таки хочеться. На честь Дня Незалежності України редакція dev.ua вирішила провести невеликий експеримент.
Ми задали чотирьом різним нейронним мережам п’ять однакових запитів: «прапор України», «День Незалежності України», «український Крим», «перемога України» та «українці». Отриманими результатами ми ділимося з вами нижче.
У TikTok тепер можна генерувати фон за допомогою нейромережі. Ми протестували її та ділимося результатами
У TikTok з’явилася нова функція «Розумний фон». З її допомогою як фон для тіктоків можна підставляти згенеровані нейромережею зображення. Редакція dev.ua протестувала цю технологію і ділиться своїми враженнями.
