OpenAI bans Russian ChatGPT accounts in influence operation

OpenAI banned Russian-linked ChatGPT accounts that promoted a suspect think tank and amplified pro-Russia narratives across social platforms.

Key takeaways

• OpenAI banned a cluster of ChatGPT accounts it assesses as very likely originating in Russia.
• The accounts used ChatGPT mainly to create promotional posts for the International Burke Institute (IBI).
• OpenAI found that 34 of 36 IBI articles reviewed had been copied from elsewhere online.
• The campaign had limited overall reach, but some associated Telegram channels had 10,000–20,000 followers.

OpenAI traces the campaign to Russian-linked accounts

OpenAI has banned a cluster of ChatGPT accounts it says very likely originated in Russia after uncovering an unusually elaborate covert influence operation built around the International Burke Institute (IBI), a self-described expert community claiming to be based in Israel.

The operators used Russian-language prompts to generate mostly English-language social posts and comments for X, Facebook, LinkedIn, Telegram and Substack. OpenAI says they instructed ChatGPT to strip out linguistic clues pointing to Russia and they used VPNs to access the service, which is unavailable there. The company says the campaign was previously unreported.

A questionable think tank supplied credibility

The AI-generated content was mainly promotional. It pushed readers toward IBI material and related channels rather than form the institute’s core articles. OpenAI says the website, registered in February 2025, presented purported experts, academic-style research and a proprietary “sovereignty” index that portrayed Russia favorably while criticizing Western countries.

That appearance of authority was fragile. In OpenAI’s review of 36 IBI articles published between September 2025 and May 2026, 34 had been copied from elsewhere online, and some were attributed to the wrong authors. OpenAI also found traces of real individuals in Israel associated with IBI activities but said it could not establish their relationship to the Russian operators.

Scalable influence playbook despite limited reach

The campaign also supported German-language Telegram activity critical of Ukraine, the EU and Germany’s government, while another operator generated logos for around a dozen channels aimed at audiences in the US, France, Poland and Türkiye. OpenAI described the infrastructure as more elaborate than Russia-linked operations it had previously disrupted.

OpenAI assessed the operation’s immediate reach as limited. Typical social posts attracted few views, although some Telegram channels had roughly 10,000 to 20,000 followers. The bigger concern is the playbook: AI can help influence actors rapidly produce polished promotion around an ecosystem engineered to look independent, credible and scalable. For readers, the case is another reason to verify who operates unfamiliar “research” organizations, inspect original sources, and treat social-media authority signals with caution.

Think before you trust or click

AI-assisted influence campaigns can borrow the visual language and authority signals of legitimate organizations, making unfamiliar sources harder to judge at a glance. Before sharing claims, look for original research, independent reporting, identifiable authors and evidence that an organization has the expertise it claims.

Suspicious posts can also lead users to deceptive or malicious links. Bitdefender Scamio can analyze questionable messages, links, screenshots and QR codes, while Bitdefender Ultimate Security adds anti-phishing, anti-fraud and AI-powered scam protection for broader protection against deceptive online content.