{"id":8353,"date":"2026-02-03T13:53:00","date_gmt":"2026-02-03T12:53:00","guid":{"rendered":"https:\/\/mediapeople.pl\/?p=8353"},"modified":"2026-05-07T12:43:53","modified_gmt":"2026-05-07T10:43:53","slug":"swarms-of-artificial-intelligence-new-front-of-information-warfare","status":"publish","type":"post","link":"https:\/\/mediapeople.pl\/en\/roje-sztucznej-inteligencji-nowy-front-wojny-informacyjnej\/","title":{"rendered":"AI swarms: a new front in information warfare"},"content":{"rendered":"

Generative AI is changing how disinformation is created and spread.<\/em><\/strong> Research published in Science highlights the growing threat of<\/em><\/strong> AI swarms: coordinated systems capable of infiltrating social media at scale,<\/em><\/strong> manipulating public opinion and undermining democratic debate.<\/em><\/strong><\/p>\n\n\n\n

Researchers warn that generative AI is creating new threats to public debate and democracy. In a commentary published in Science<\/em> they describe AI swarms: sophisticated multi-agent systems that can infiltrate social media at scale, imitate human behaviour, manipulate emotions and amplify selected narratives.<\/p>\n\n\n\n

Traditional bots usually carry out simple, repetitive tasks. AI swarms, by contrast, can be coordinated by large language models (LLMs) and operate as a self-sustaining system. They can learn in real time, adapt to a platform and its users, and evade moderation. This represents a qualitative shift towards distributed influence systems capable of holding conversations and building credibility.<\/p>\n\n\n\n

From bots to malicious AI swarms<\/strong><\/p>\n\n\n\n

The authors writing in Science<\/em> argue that advances in generative language models and multi-agent systems make malicious AI swarms possible. These autonomous, coordinated groups of agents can act as an adaptive collective, responding to their surroundings while maintaining distinct identities and memories. They can tailor content to different audiences and operate across several platforms simultaneously over extended periods.<\/p>\n\n\n\n

As the authors of the publication note, it is these capabilities that create a new category of threat:
\u201cThese capabilities create a destabilising threat: swarms of cooperating, malicious AI agents. By adaptively imitating human social dynamics, these systems threaten democracy.\u201d<\/p>\n\n\n\n

Manipulation through synthetic consensus<\/strong><\/p>\n\n\n\n

The risk goes beyond producing false content. The publication identifies a deeper problem: manipulating social structures by creating synthetic consensus \u2014 the impression that everyone shares an opinion, even when they do not. Influence here depends on shaping perceptions of majority opinion rather than persuading people through arguments.<\/p>\n\n\n\n

AI swarms could therefore influence social norms, beliefs and decisions, weakening rational debate and trust in independent voices. As the researchers put it:
\u201cThe main threat is not false content alone. It is synthetic consensus: the illusion of widespread agreement, manufactured at scale.\u201d<\/p>\n\n\n\n

How AI swarms work<\/strong><\/p>\n\n\n\n

The authors define an AI swarm as a distributed network of AI-controlled agents working towards shared goals. The agents can maintain persistent identities and remember relationships and previous interactions, making their presence in a community appear more credible than random spam.<\/p>\n\n\n\n

A swarm can coordinate around a shared goal while varying the tone, content and roles of individual accounts to resemble the diversity of real users. It observes platform signals and audience reactions, learning and refining its strategies with minimal human oversight. The authors define it as follows:
\u201cA malicious AI swarm is a network of AI-controlled agents that can maintain persistent identities and memories, coordinate around shared goals while varying the tone and content of messages, and adapt to user engagement and reactions.\u201d<\/p>\n\n\n\n

Scale and impact: from manipulation to harassment<\/strong><\/p>\n\n\n\n

The scale of an AI swarm depends on available computing resources and the safeguards built into a platform. Although discussions often refer to hundreds of thousands or millions of agents, the authors argue that credibility and adaptability matter as much as numbers. Even a relatively small swarm can create pressure to conform and influence a debate.<\/p>\n\n\n\n

The publication describes several potential harms. AI swarms can manipulate public opinion, influence democratic processes, weaken independent critical debate and fragment the shared information environment, particularly in already polarised societies. Their uses may range from commercial influence to political campaigns and state operations.<\/p>\n\n\n\n

As they conclude:
\u201cAdvances in AI make it possible to manipulate beliefs and behaviour across entire populations. Generative tools allow propaganda to scale without losing credibility.\u201d<\/p>\n\n\n\n

In practice, AI swarms can combine two forms of influence: distorting perceptions of what is widely accepted and exerting social pressure. That pressure may involve mass harassment, the appearance of an angry crowd or attacks on people expressing different views, discouraging them from joining the debate or driving them off a platform.<\/p>\n\n\n\n

Proposed safeguards<\/strong><\/p>\n\n\n\n

The authors argue that deleting individual accounts and reacting to harmful content will not be enough. Because swarms are adaptive and distributed, the response must combine several layers of technical and institutional safeguards.<\/p>\n\n\n\n

They call for detection methods based on behavioural analysis, including unusual patterns of coordination that may reveal a swarm. Stronger identity checks and proof-of-personhood tools are another priority. The authors also propose simulations and resilience testing to anticipate how swarm behaviour may evolve, supported by an AI Influence Observatory: a global network monitoring and analysing AI-driven influence operations.<\/p>\n\n\n\n

As the researchers emphasise: \u201cDefences against these systems must be layered and pragmatic, increasing the cost, risk and visibility of manipulation.\u201d<\/p>\n\n\n\n

Implications for public debate and businesses<\/strong><\/p>\n\n\n\n

If AI swarms operate at scale, pressure on platforms, institutions and organisations will increase. For democracy, the risks include declining trust, reduced pluralism and rational debate giving way to emotional pressure and an artificial sense of majority opinion. For brands and businesses, swarms could create reputational crises and make it harder to distinguish genuine public sentiment from coordinated influence.<\/p>\n\n\n\n

As the boundary between human and AI-generated activity becomes less clear, resilience to information manipulation is becoming a strategic priority. AI swarms thrive in confusion. The strongest response is to keep improving the credibility, transparency and quality of debate in the spaces for which we are responsible.<\/p>\n\n\n\n

References<\/strong><\/p>\n\n\n\n

    \n
  1. Science<\/strong>\u00a0(2026).\u00a0How malicious AI swarms can threaten democracy<\/em>. Policy Forum.<\/li>\n\n\n\n
  2. Live Science<\/strong>\u00a0(2026).\u00a0Next-generation AI swarms will invade social media by mimicking human behaviour and harassing real users, researchers warn<\/em>.
    https:\/\/www.livescience.com\/technology\/artificial-intelligence\/next-generation-ai-swarms-will-invade-social-media-by-mimicking-human-behavior-and-harassing-real-users-researchers-warn<\/a><\/li>\n\n\n\n
  3. EurekAlert!<\/strong>\u00a0(2026).\u00a0AI swarms could quietly distort democracy, science policy forum warns<\/em>.<\/li>\n\n\n\n
  4. City St George's, University of London<\/strong>\u00a0(2026).\u00a0AI swarms could fake public consensus and undermine democracy<\/em>.<\/li>\n<\/ol>","protected":false},"excerpt":{"rendered":"

    Rozw\u00f3j generatywnej sztucznej inteligencji zmienia spos\u00f3b, w jaki powstaje i rozprzestrzenia si\u0119 dezinformacja. Najnowsze badania opublikowane w czasopi\u015bmie Science wskazuj\u0105 na rosn\u0105ce zagro\u017cenie ze strony tzw. roj\u00f3w AI – skoordynowanych system\u00f3w sztucznej inteligencji zdolnych do masowej infiltracji medi\u00f3w spo\u0142eczno\u015bciowych, manipulowania opini\u0105 publiczn\u0105 i podwa\u017cania fundament\u00f3w demokratycznej debaty. Naukowcy alarmuj\u0105, \u017ce rozw\u00f3j generatywnej sztucznej inteligencji otwiera […]<\/p>\n","protected":false},"author":1,"featured_media":8354,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_seopress_titles_title":"","_seopress_titles_desc":"","_seopress_robots_index":"","_seopress_robots_follow":"","_seopress_robots_imageindex":"","_seopress_robots_snippet":"","_seopress_robots_primary_cat":"","_seopress_robots_breadcrumbs":"","_seopress_robots_freeze_modified_date":"","_seopress_robots_custom_modified_date":"","_seopress_robots_canonical":"","_seopress_social_fb_title":"","_seopress_social_fb_desc":"","_seopress_social_fb_img":"","_seopress_social_fb_img_attachment_id":0,"_seopress_social_fb_img_width":0,"_seopress_social_fb_img_height":0,"_seopress_social_twitter_title":"","_seopress_social_twitter_desc":"","_seopress_social_twitter_img":"","_seopress_social_twitter_img_attachment_id":0,"_seopress_social_twitter_img_width":0,"_seopress_social_twitter_img_height":0,"_seopress_redirections_value":"","_seopress_redirections_enabled":"","_seopress_redirections_enabled_regex":"","_seopress_redirections_logged_status":"","_seopress_redirections_param":"","_seopress_redirections_type":0,"_seopress_analysis_target_kw":"","inline_featured_image":false,"footnotes":""},"categories":[17],"tags":[],"class_list":["post-8353","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-aktualnosci"],"_links":{"self":[{"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/posts\/8353","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/comments?post=8353"}],"version-history":[{"count":1,"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/posts\/8353\/revisions"}],"predecessor-version":[{"id":8355,"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/posts\/8353\/revisions\/8355"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/media\/8354"}],"wp:attachment":[{"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/media?parent=8353"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/categories?post=8353"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/mediapeople.pl\/en\/wp-json\/wp\/v2\/tags?post=8353"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}