The Integrity of Information: An Unfeasible Project?
World Press Freedom Day and the International Agenda
World Press Freedom Day was celebrated on May 5th. This event was established by the United Nations General Assembly in 1993 to draw attention to the central role of the press in building stronger democracies. Globally, the commemorations were led by UNESCO, with a conference held this year in Brussels under the theme "Reporting in the Brave New World: The Impact of Artificial Intelligence on Press Freedom and the Media."
During the event, various working papers were presented, including “Freedom of Expression, Artificial Intelligence and Elections" a brief report prepared by electoral expert Ajay Patel, commissioned by UNESCO and the United Nations Development Programme (UNDP). The report aims to provide an overview of the impact of Artificial Intelligence (AI) on the exercise of freedom of expression during electoral processes, offering analytical insights to help identify lines of action tailored to different contexts.
A Blurred Framework: Critiques of the “Information Integrity” Approach
While the report offers various examples of AI use in electoral contexts—both positive and negative—it lacks a deeper reflection from a freedom of expression perspective. Unfortunately, this has been a recurring issue in many efforts led by the United Nations (UN) and its affiliated bodies under the umbrella of the "information integrity" agenda. The clearest example of this is that, to date, the UN has yet to provide a solid justification for treating fundamentally different phenomena, such as disinformation and hate speech, under the same analytical framework. Although there may be some similarities and points of overlap between the two, they are inherently distinct issues, even in terms of how they are addressed legally according to international standards governing restrictions on the flow of information. What is gained and what is lost by framing them as part of the same problem remains a question that has yet to receive a clear and convincing answer.
Perhaps a brief report is not the most appropriate venue to address this gap, and this is understandable. However, the lack of a substantive, thoughtful examination that weighs the pros and cons of adopting this analytical framework in relation to the full exercise of freedom of expression undermines the very interests of the United Nations and its agenda. It seems that the topic is being deliberately avoided in a series of consistently introductory reports, more focused on promoting terminology than on fully understanding the consequences of such a framework.
Another example of this is the lack of clarity regarding the level at which threats and potential solutions should be analyzed. At times, the report appears to advocate for a systemic understanding of electoral contexts, contexts involving diverse actors playing different roles over a defined period. It also acknowledges that not all "inaccurate" information is necessarily "harmful" and that trust in an election is not determined solely by its "technical qualities" but rather by a broader political context. Yet, when identifying potentially harmful uses of AI in electoral contexts, the analysis is limited to the most granular level: specific messages and individual units of information.
The problem is that if the mere existence of a "false" message—whether constructed with artificial intelligence or not and disseminated through social media—is considered a threat to information integrity, then information integrity becomes an unachievable project. Setting aside the type of measures that would be necessary to attempt such a goal, the implications for freedom of expression, and the questionable desirability of such a pursuit within a democratic framework, the bottom line is simple: it’s an unattainable objective. So, what is the use of an analytical framework built on that premise? How is the success of policies aimed at achieving that goal supposed to be measured?
This is not about denying that certain misinformation and malicious misinformation can be harmful, nor about surrendering or remaining passive in the face of potential harm, but about accepting that social life is not reduced to a series of factual data. This is especially true in electoral contexts, which are necessarily crisscrossed by desires, promises, beliefs, trusts, interpretations, opinions, strategies, offensives, appearances, disappointments, rivalries, and affections, separated from the facts by a boundary that is often tenuous. And all of this is part of life in a democracy, which is itself often a yearning and a contested territory.
Citing the report of the United Nations Secretary-General, Strengthening the role of the United Nations in enhancing periodic and genuine elections and the promotion of democratization, policy brief states that “false and misleading” information can “amplify tensions and divisions” by undermining the electorate's ability to make informed decisions. The General Secretariat's report adds that there are no simple technical solutions to counter the effects of disinformation, which is true. But even more so, it is worth asking whether the problem is really technical in nature; perhaps this analysis has inverted the cause-effect relationship, and disinformation is merely a symptom of a larger sociopolitical problem, a consequence of a context where the conditions for informed elections have been previously eroded by a lack of legitimacy of the institutions called upon to guarantee them, including political parties, the press, the government, and the bodies responsible for overseeing electoral processes.
Clearly, the proliferation of discourse that exacerbates these tensions is unhelpful, but focusing the response solely on such expressions fails to offer real solutions—and, in the long run, may even make things worse. If the institutions responsible for shaping and implementing public policy lack the necessary legitimacy to be perceived as appropriate and necessary, their actions may actually increase polarization in already unstable contexts. If that is the case, then the kind of solutions required should focus less on pursuing and correcting malicious information and more on rebuilding institutional legitimacy. Naturally, this type of solution is harder to implement, takes longer, and is likely less attention-grabbing.
One striking element appears on page 7 of the document: a diagram intended to visually represent the “Information Environment.” It includes columns for state actors, the public, the media and platforms, political actors, and civil society. Hovering above them all are“Malicious Actors / Hackers”and“Foreign State Influence Actors”The diagram is neither explained nor discussed in the report. What stands out is that the framework closely resembles one that might be generated from a military or national security perspective, implying that the main threats are external and situated in an international conflict scenario. It is questionable, first, whether such a framework—where disinformation appears as a Cold War-era leftover—can be considered the most appropriate model outside of the United States and Europe.
And What About Latin America?
Countries with fragile democracies and extensive histories of authoritarianism, repression, and dictatorship—like those in Latin America—have been forced to develop a certain legitimate, and at times useful, distrust toward state institutions and major media conglomerates. In Chile, El Mercurio and Copesa, the two largest newspaper publishing companies in the country to this day, were active accomplices of the Pinochet dictatorship, using their pages to cover up crimes against humanity. The lack of media pluralism in such contexts represents a far deeper problem for democracy than any isolated use of generative artificial intelligence systems.
This is not to say that the solutions presented in the brief report are inherently bad. The various documents that the United Nations has produced under its agenda of Information Integrity have consistently emphasized, on the one hand, the difficulty of balancing freedom of expression with punitive measures. On the other hand, they promote actions such as media and information literacy, ensuring the free and safe practice of journalism, and encouraging multistakeholder decision-making processes. All of these can be positive and desirable measures. However, without a clear understanding of both the advantages and disadvantages of adopting the “information integrity” framework—beginning with a precise definition of the term—it is difficult to determine whether such measures are capable of achieving the intended outcomes.
What does information integrity actually look like? Is it a binary concept? How is it defined? Based on the raw volume of false information circulating? Proportionally, in relation to the amount of true versus false information? By the impact false information generates? How is that impact measured? How do we know which areas of the information ecosystem need to be prioritized? How do we assess the differentiated impact that various actors have within a specific context? All of these are questions that, for now, remain unanswered.
The landscape is complex. Information Integrity has become a key term in the global discussions around internet governance, despite its vague meaning and still-weak theoretical foundation. Before continuing to promote its adoption in an increasing number of areas, it would be wise for the United Nations to address the fundamental questions this approach raises, to prevent these efforts from collapsing in the near future due to a lack of solid groundwork.
