OpenAI banned accounts linked to multiple Russia-origin AI-assisted influence operations (“No Bell,” “Fish Food,” and “Trolling Stone”).
Security issue ● Confirmed 86% confidence first seen
OpenAI reported taking action against accounts tied to three previously unreported influence efforts that it says originated from Russia and used AI for targeted online commentary across different platforms and languages. The bans removed access to the associated accounts and were intended to disrupt their AI-enabled influence activity.
Decision brief
- What changed
- OpenAI said it banned accounts tied to three previously unreported AI-assisted influence operations it linked to Russia-origin or Russia-linked actors: “No Bell,” “Fish Food,” and “Trolling Stone.” According to OpenAI, the bans removed those accounts’ access to its services and disrupted their ability to use OpenAI tools for targeted multilingual commentary and online influence activity.
- Why it matters
- For business leaders, this is a concrete example that general-purpose AI services are being used in coordinated influence operations across regions, languages, and platforms, which raises the operational importance of abuse monitoring, platform controls, and incident response. It also shows that AI providers may take disruptive enforcement action against suspected influence networks, which matters for organizations that rely on third-party AI platforms and need clear governance for acceptable use, trust-and-safety, and reputational risk.
- Evidence
- The only cited sources are three OpenAI blog posts, each describing one operation and stating that OpenAI banned the related accounts: “No Bell,” “Fish Food,” and “Trolling Stone.” Across the three posts, OpenAI consistently says the activity was Russia-origin or Russia-linked and that the enforcement action removed access and disrupted AI-enabled influence efforts, but the coverage provided does not include independent external confirmation.
- What remains uncertain
- Key details remain unverified from the provided coverage, including the full attribution basis for linking these operations to Russia, the scale and impact of the campaigns, and whether any activity continued using other tools or providers after the bans. It is also unclear what detection methods OpenAI used, how quickly it identified the abuse, and whether similar influence networks remain active on other AI platforms.
- Monitor next
- Watch for independent reporting or government/platform disclosures that corroborate the attribution, scope, and downstream disruption effects of the three banned operations.
Analytical support, not advice — assumptions and open questions stated above.