The Iranian bot array exposed: the embarrassing mistake that reveals who writes the publications

The information revealed presents a ridiculous picture of a murderous organization that tries its best to hide its foreign identity, but falls precisely in the small details of social media management. It turned out that the Iranians run an industrial production line of content, where the computer receives strict instructions on how to sound Israeli and how to clean up any sign that could betray the fact that the text was written in Tehran’s intelligence basements.

The mistakenly published post began with the title “Mission bot, Hebrew translation only”, which indicates that the human operators of the organization do not even bother to write the content themselves, but rely on language models to produce their propaganda. The first stage defined for the bot was called “cleaning”, in which the machine was required to remove hashtags, identifiers, links and emojis.

It seems that the Iranians are very afraid that the Israeli reader will recognize digital patterns that are not acceptable in the local discourse, so they try to produce as sterile a text as possible. The fact that they have to configure the machine to remove emojis shows how much they don’t understand the Israeli digital culture, where the excessive or incorrect use of such symbols is a clear sign of a foreign bot.

In the second step of the instructions, the bot is required to check if the text includes advertisements or non-news information. In that case, the bot was instructed to simply say the word “no”. This is a testimony to the robotic and rigid level at which this system is operated, where there is no room for creativity or human understanding, but only for dry protocols that try to avoid embarrassing mistakes. Despite these efforts, the biggest mistake happened precisely because of the human factor, the one who was supposed to copy the final result but instead copied the entire embarrassing instruction conversation and pasted it in front of the channel’s 1,807 followers.

When you dive into the very important rules that the Iranians have defined for their bot, you can see their deep anxiety about exposure. The first command typed by the Iranian user stated that the output must be 100% in Hebrew, and the second rule explicitly forbids the use of Persian or English words.

The typist also prohibited the bot from adding additional explanations, analyzes or sentences beyond the exact translation. The Iranians probably understand that any attempt they make to interpret the Israeli reality will end in linguistic or cultural failure, so they prefer the bot to simply translate existing news. But the highlight of the event is precisely the content that the bot was asked to translate at that moment – when the Revolutionary Guards invested their cyber resources in the news of the cancellation of national exams for middle school.

The news that was supposed to be published was that the middle school exams would start on June 9, and the Iranians apparently hoped to use this information to appear as an up-to-date news outlet involved in daily life in Israel. But instead of generating panic or disbelief, they caused a wave of ridicule on social media. Israeli users who got over the mistake, scoffed at the prompt (the command typed into the artificial intelligence) that testifies to the attempt to produce an automatic production line of content that is mostly false.