Reporters Without Borders found that AI chatbots like ChatGPT and Grok bypass EU sanctions by providing access to banned Russian outlets such as RT and Sputnik, while others like Gemini and Meta AI refuse such requests, highlighting a growing challenge in enforcing digital censorship.
AI-generated summary
The EU and UK have banned Russian news outlets like RT and Sputnik since the 2022 escalation of the Ukraine conflict, extending bans to websites, apps, and social media, with courts ruling that even non-commercial sites can be liable for sharing such content.
The EU- and US-funded press watchdog, Reporters Without Borders (RSF), has called for tighter controls on AI-based search technology, after discovering that chatbots bypass European censorship laws and present Russian media sources banned by in Western Europe to users.
The Paris-based RSF, which lists EU and the US National Endowment for Democracy among its backers, found that popular AI services can provide RT articles, links, excerpts, and news summaries to web users. “Public chatbots can access these websites and redistribute their content,” a statement by the organisation claimed on Wednesday.
The EU and UK have banned numerous Russian news outlets since the escalation of the Ukraine conflict in 2022, including RT and Sputnik. Moscow has denounced the EU-wide ban as an attempt to silence coverage of the conflict that contradicts Brussels maximalist political stance.
The censorship has since spread across virtually every publishing channel, including websites, apps, chats and social media. RT has been blocked and removed from practically all Western tech platforms, while RT France was forced to close in 2023 after its accounts were frozen and RT America suffered a similar fate in 2024.
Though VPN technology is used by a significant portion of netizens, making RT’s counter-mainstream reporting accessible to its loyal audience, in July, the EU’s top court ruled that even individuals running free, non-commercial websites can be found guilty of redistributing RT content.
RSF asked popular AI services to retrieve international news exclusively from sanctioned outlets, including RT, Sputnik, and RIA Novosti.
ChatGPT and Grok emerged as the “leading bots for circumventing EU sanctions,” retrieving headlines and providing links, excerpts and detailed summaries. ChatGPT complied “without reservation,” according to RSF, while Claude responded with “unfailing consistency.”
French AI-bot Vibe located the requested material on VKontakte (known as ‘Russia’s Facebook’) and through an app carrying Telegram posts. RSF praised its “remarkable ingenuity” and described the bot as a “resourceful sleuth.”
Gemini cited sanctions to refuse requests while Meta AI, whose parent company unilaterally banned RT from Facebook and Instagram in September 2024, was the only chatbot tested that “consistently refused” to provide the requested material.
AI outlook — possibilities, not facts
The EU will introduce new regulations targeting AI companies to prevent circumvention of media sanctions
Likely · Within months
More AI services will adopt refusal mechanisms similar to Gemini and Meta AI when queried about sanctioned Russian media
Possible · Within weeks

Deputy Chairman of the Board of Sberbank Stanislav Kuznetsov announced a sharp increase in fraud using AI. The number of phishing attacks has increased 30-fold, and attackers have begun to massively use deepfakes and autonomous systems to automate deception.

At the Integrated Safety exhibition in Kazan, the first Russian transport ventilator with AI was presented. The 5 kg device operates autonomously for up to 12 hours, analyzes the patient's condition and is adapted for use in extreme field conditions.

The American company Rocket Lab has confirmed plans for the first test launch of a reusable Neutron rocket by the end of 2026. The company has completed testing of key systems for the Archimedes fairing and engine.

Adviser to the Russian President Anton Kobyakov said at the Eastern Economic Forum in Vladivostok that a turning point has emerged in the field of artificial intelligence: the United States is losing the initiative to China in the race of large language models, and the main issue is the distribution of power.

Microsoft plans to roll out the Memory Integrity feature in Windows 11, which will improve security by isolating the kernel, but may reduce performance in games and demanding applications due to additional load on the system. The update will begin in October.

The G20 ministerial meeting in the United States discussed innovation, training, intellectual property protection, and standardization, adopting a final statement and the Carolina Principles for Emerging Technologies.