OpenAI has banned accounts linked to a Russia-origin operation, dubbed "Bad Grammar", that used AI models to generate English- and Russian-language Telegram comments. This network targeted audiences in Russia, Ukraine, the United States, Moldova, and the Baltic States.

The "Bad Grammar" operation aimed to automate comment-spamming on Telegram. Operators used OpenAI's models to debug posting code and generate comments on political topics such as the war in Ukraine, and politics in Moldova, the Baltic States, and the United States. In English, fake personas were used to represent different political viewpoints. In Russian, comments were more thematic.

The network primarily focused on a few Telegram channels, including @Slavyangrad, @police_frequency, and @SGTNews Network. Content often accused leaders of corruption and argued against US support for Ukraine. Following the March 22 Moscow attack, new accounts emerged focusing on Tajikistan, calling for unity and respect for Russian law.

While using novel AI techniques, the operation's impact was limited, with few comments receiving engagement. OpenAI assessed its impact as Category 1 on the Breakout Scale, indicating single-platform activity without significant amplification.