AI swarms could be the next evolution of disinformation. Instead of individual bots, networked AI systems act in a coordinated manner and thus simulate social consensus. Researchers warn that this development could significantly endanger public discourse and democratic decision-making processes.
Artificial intelligence is driving digital transformation forward enormously and is already shaping key areas in business, science and society. In a very short time, AI has already changed how information is created, distributed and perceived.
The ability to create and coordinate content at scale opens up new opportunities. But it also entails significant risks. AI is increasingly developing into a power factor that can influence opinion formation and political processes.
Researchers therefore urgently warn about the social consequences of this development. They see AI swarms as the next generation of disinformation because they appear in a coordinated manner, deceptively imitate human debates and can thus deliberately distort opinions.
Are AI swarms endangering democracy?
An international consortium of researchers warns of a new level of escalation in digital manipulation. So-called AI swarms could be used specifically in the 2028 US presidential election to massively disrupt the electoral process.
Leading AI researchers and social scientists from Berkeley, Harvard, Oxford, Cambridge, Yale and Konstanz, among others, have identified AI swarms as a new, hard-to-detect form of coordinated disinformation. These have the potential to specifically manipulate the formation of public opinion and undermine democratic processes.
These systems are able to coordinate autonomously, infiltrate communities and establish consensus efficiently. By adaptively mimicking human social dynamics, they threaten democracy.
In their article in the renowned specialist magazine Science the researchers describe this “disruptive threat” posed by “swarms of collaborative, malicious AI agents.” This threatens to create an unprecedented form of opinion manipulation that seriously endangers democratic discourse.
Is there a risk of election manipulation by AI swarms?
The research team warns above all about the “artificial consensus” that could arise from the AI swarms. This is because it is able to influence beliefs and norms – even with controversial claims.
“The danger is no longer just in fake news,” explains Jonas R. Kunst from the BI Norwegian Business School, one of the main authors of the article. Rather, it is dangerous “that the foundation of democratic discourse – independent voices – collapses when a single actor can control thousands of unique, AI-generated profiles.”
In addition, harmful AI swarms would have a negative impact on the training data of regular AI systems. As they flood the internet with fake claims, they could expand their influence on established AI systems.
According to the researchers, this is not just a theoretical threat. Analyzes have shown that tactics in this form are already being used. They were used in the 2024 elections in Taiwan, India and Indonesia.
The researchers are therefore calling for coordinated measures worldwide to contain this risk. They suggest so-called “swarm scanners” and watermarked content. In this way, AI-controlled disinformation campaigns could be counteracted.
“Beyond the deception or security of individual chatbots, we need to explore new threats that arise from the interaction of many AI actors,” explains David Garcia, Professor of Social and Behavioral Data Science at the University of Konstanz. “It is essential that we study these AI actors using behavioral science methods and analyze their collective behavior when they interact in large groups.”
Also interesting: