UN Warned: AI-Powered Disinformation Threatens Information Integrity as General Debate Focuses on Digital Manipulation
UNITED NATIONS — During the fourth and fifth days of the UN General Assembly’s General Debate, a series of speakers turned their attention to a rapidly evolving threat: the growing ability of digital tools to amplify false information on an unprecedented scale. While much of the debate focused on familiar diplomatic concerns, a growing number of countries warned that information integrity, once viewed largely as a domestic media issue, is now a global security challenge. The central concern, speakers said, is no longer only the creation of false content, but the capacity of networked platforms, automated systems and artificial intelligence to distribute falsehoods to billions of people, shape public debate and undermine trust in democratic institutions. The warnings reflected a deepening unease among member states about the ways in which the same technologies that connect humanity can be turned into weapons against it. From small nations to larger powers, the message was consistent: the machinery of information has become a battleground, and artificial intelligence has raised the stakes dramatically. The General Debate, which brings together world leaders at UN headquarters, has become an increasingly important venue for discussing digital policy, and the interventions on Days 4 and 5 made clear that disinformation is no longer seen as an incidental problem but as a core threat to sovereignty, democracy and fundamental rights. Speeches ranged from detailed national proposals to broad calls for international cooperation, but all shared a common assumption: the information environment is being fundamentally altered by AI, and existing governance frameworks have not kept pace.
Albania offered one of the most detailed interventions on this issue, arguing that the traditional understanding of freedom of speech is being overtaken by what it called “freedom of reach.” In a digital environment, Albania said, a lie can be multiplied a billion times, making censorship unnecessary. Instead of needing to silence truthful voices, those who seek to manipulate public opinion can simply bury the truth beneath an avalanche of falsehoods. Albania argued that this reach is not accidental or neutral; it can be purchased, automated, manipulated and secretly assigned by platforms, corporations, criminal networks and foreign states. This insight reframes the policy problem. The right to speak is meaningful only if the mechanisms that determine who hears that speech are transparent, fair and resistant to abuse. Turning to artificial intelligence, Albania warned that the greatest immediate danger is not that AI systems will someday stop obeying humanity, but that they could “obey the wrong human beings perfectly.” In other words, the risk is not rogue machines but human misuse. AI has the capacity to multiply existing human prejudices and can be deployed by states to spread false information on an enormous scale. In this sense, AI is not a neutral tool; it is an amplifier of human choices, including the choice to manipulate. Albania’s framing suggested that the international community has been too focused on the hypothetical risks of superintelligent AI while underestimating the very real and present danger that existing AI systems will be used by bad actors to distort public discourse and undermine democratic decision-making.
Iceland echoed those concerns and brought a specific focus on the manipulation of democratic processes. It argued that AI is making it easier to produce convincing falsehoods and to manipulate public debate, creating risks that cut across national borders and affect all societies. Foreign states and other actors, Iceland said, must not be allowed to use AI to spread falsehoods, manipulate democratic processes or deceive citizens at unprecedented speed and scale. The speed of AI-generated content, combined with its low cost, gives malicious actors an advantage over journalists, regulators and ordinary citizens trying to verify what they see online. Iceland called for international cooperation between governments, scientists and businesses to harness AI’s benefits while confronting its risks. This emphasis on multi-stakeholder cooperation reflected a recognition that no single country, however powerful, can solve the problem alone. North Macedonia placed those concerns within a broader information environment, warning that people have never before been exposed to so much manipulation, fabricated images, propaganda and organised disinformation. That observation underscores the scale and persistence of the problem: manipulated content is no longer an occasional anomaly but a constant feature of online life in many countries, and citizens are increasingly unable to tell the difference between authentic information and coordinated falsehoods. Gabon, speaking from the perspective of a country concerned with stability and fundamental rights, identified disinformation, misinformation and information manipulation as threats to the security of states and the rights of citizens. It said it had introduced temporary and proportionate regulation of the digital and AI sectors with the stated aim of protecting citizens, demonstrating that even countries with limited technological resources are being forced to respond to the new information landscape.
Albania did not stop at diagnosis. It outlined a series of national and international measures intended to counter manipulation and restore a measure of transparency to the digital public square. At the national level, the country said it is building an AI-based “digital mirror” designed to identify manipulation in real time, test claims against authoritative information, reveal coordinated patterns and help citizens understand not only what they are seeing online, but how and why it reached them. The digital mirror represents an attempt to use AI defensively, matching the speed and scale of automated manipulation with automated verification and transparency. It reflects a broader shift from reactive fact-checking to continuous, systems-level monitoring of the information environment. Such a tool would not simply tell citizens whether a piece of content is true or false; it would show the paths through which content travels, the accounts and networks that amplify it, and the techniques used to increase its visibility. At the international level, Albania called for an international compact for sovereign and democratic AI. Under that compact, citizens should know when politically consequential content has been artificially generated or materially manipulated. This is a form of algorithmic transparency that goes beyond existing labelling requirements, which often apply only to obvious deepfakes and can be easily avoided. Albania also called for greater transparency around how consequential political content is created, financed, targeted and artificially amplified, and it proposed a shared early warning system through which evidence of AI-driven manipulation could move rapidly between participating countries. That system would allow governments and institutions to respond before a lie solidifies into a false narrative that shapes elections, public health decisions or international relations.
Taken together, the speeches highlighted a significant shift in concerns about online disinformation: from the creation of false content alone toward the systems that can distribute and amplify it at scale. This distinction matters because the simple act of publishing a falsehood has been possible for centuries. What has changed is the ability to ensure that the falsehood is seen, believed and acted upon by precisely the people most likely to be swayed. Albania’s distinction between freedom of speech and “freedom of reach” focused attention on that difference. The issue is not simply whether someone can publish false information, but how platforms, automation and other actors can dramatically increase its visibility and influence. The algorithms that decide what appears in news feeds, the advertising systems that target users by psychological profile, and the networks of fake accounts that create the illusion of popularity are all part of this distribution system. AI affects both sides of the equation. As Iceland noted, it can make convincing falsehoods easier to produce, generating text, images, audio and video that look authentic and can be personalised for individual targets. At the same time, as Albania warned, AI can multiply and distribute manipulation on an enormous scale. It can automate the operation of bot networks, coordinate the timing of messaging and adapt to platform defences in ways that human-run campaigns cannot. The combination is particularly dangerous: cheap, abundant falsehoods combined with powerful distribution engines create an information environment in which even the most sophisticated citizens can find themselves deceived.
That creates fundamental challenges for information integrity and democratic governance. The volume and speed of manipulated content can make it more difficult for citizens to determine what information is authentic, and when uncertainty becomes pervasive, public trust erodes. In an environment where a fabricated image can circulate around the world before verification is possible, trust in institutions, media and even personal communication weakens. North Macedonia’s warning about fabricated images, propaganda and organised disinformation reflected the broader information environment in which AI-related risks are emerging. The responses discussed during the UN General Assembly’s General Debate ranged from national tools for identifying manipulation to calls for transparency around artificially generated political content and international information-sharing mechanisms. Those speeches did not set out a single approach to addressing disinformation. They did, however, illustrate how the discussion is expanding beyond the content of individual messages to include the technologies and systems that determine how information is created, amplified and delivered to citizens. The emerging consensus is that defensive measures must be as sophisticated as the attacks themselves. Whether through AI-enabled digital mirrors, international early warning systems or transparency requirements for political content, governments are beginning to recognise that protecting information integrity requires not just regulating speech, but governing the infrastructure of communication. The debate is likely to continue in the coming months and years, but the warnings from Days 4 and 5 made one thing clear: the era of information as a weapon has entered a new, AI-powered phase, and the international community is only beginning to understand the scale of the challenge.