In 2026, the new political propaganda does not need a face, a name, or a team of people sitting at computers. All it needs is a network of accounts that look like ordinary users. They comment on elections, respond to posts, argue, amplify emotions, and create the impression that a particular opinion has suddenly begun to dominate. Increasingly often, those voices may not belong to people at all, but to artificial intelligence systems<\/strong>.<\/p>\n\n\n\n AI bots in social media no longer have to simply repeat a propaganda message. They can hold a conversation, shift tone, respond to arguments, and adapt their message to a particular community<\/strong>. That is why researchers speak of a new risk: not only disinformation online, but the production of a false sense of social consensus.<\/p>\n\n\n\n This digital identity may soon come to dominate the internet. It can not only impersonate real users, but also shape opinions by seemingly speaking through the comments of thousands of internet users<\/strong>. Each voice sounds slightly different, yet each pushes the recipient toward the same message. Before our eyes, a new form of political propaganda is emerging, and it is one of the forms of online manipulation.<\/p>\n\n\n\n This is exactly the danger highlighted by the authors of an article published in Science<\/a><\/em>. As the article explains, AI bots in social media are far more dangerous than the bots that have long been present online. They stand out because of their extraordinary speed in responding to feedback<\/strong>. They can coordinate instantly and maintain a coherent narrative, creating what researchers call a synthetic crowd and spreading disinformation online.<\/p>\n\n\n\n All of this will become possible because of the development of large language models and multi-agent systems capable of managing vast networks of AI \u201cvoices.\u201d Although algorithms<\/a> create them, they can resemble real people with uncanny precision<\/strong>. They can imitate tone and local language, and they can also interact naturally with real users. This makes an AI bot so human-like that exposing it becomes difficult. And that is not all.<\/p>\n\n\n\n An AI bot in social media can refine its conversation strategy with people<\/strong> in real time and create a substitute for broad social agreement. In reality, artificial intelligence activity can be designed specifically to influence political conversations.<\/strong><\/p>\n\n\n\n Although the full potential of AI personas still belongs partly to the future, the real threat posed by artificial intelligence is already affecting the world. Deepfake<\/a> materials and fake news sites created by language models offer one example<\/strong>. They have influenced election discussions in countries including India, the United States, Indonesia, and Taiwan.<\/p>\n\n\n\n The problem has also appeared in Europe. According to organizations that monitor the internet, pro-Russian networks have emerged and are spreading enormous volumes of content. Such activity may influence the data used to train AI models deployed in politics, and in turn affect the way these systems prioritize information.<\/p>\n\n\n\nAI in Politics Creates a Crowd Without People<\/h3>\n\n\n\n
A False Majority Sounds Convincing<\/h3>\n\n\n\n
An Army of AI Bots Is Taking Over Social Media<\/h2>\n\n\n\n
AI in Politics Is Already Shaping Electoral Debate<\/h2>\n\n\n\n