The world today has changed not only technologically but also fundamentally in terms of information. People no longer search for news—it finds them. The TikTok algorithm feeds us “truth” in one-and-a-half-minute videos, while Facebook already “knows” what interests us even before we realize it ourselves. All of this is convenient and fast, but at the same time—very dangerous. Because in the era of artificial intelligence, deepfakes, and self-learning algorithms, information is transformed from fact into a virtual construct. And with each passing day, it becomes increasingly difficult to distinguish where truth ends and manipulation begins.
At first glance, artificial intelligence (AI), algorithms, and media are simply tools. But in practice, they determine what we see, what we think, whom we trust, and even whom we vote for. In the 20th century, propaganda required printing presses and an army of censors. Today, a few lines of code, a deep neural network, and thousands of bots are enough to create a mass illusion of truth. This is the modern battlefield—without explosions, but with a devastating impact.
And today, this is not just a question for cybersecurity specialists or political analysts. It concerns all of us, because even if we don’t create disinformation—we read it, share it, and sometimes unintentionally help it spread. And if we don’t start understanding the mechanisms behind its operation, we will surely lose the battle for consciousness and critical thinking.
American society has already experienced the force of information attacks—from election campaigns to COVID disinformation. However, Ukrainians have felt it even more—because for us, this is part of an ongoing hybrid war. That is why it is important to look at this issue comprehensively: how it works technologically, who uses it and for what purpose, and what we as citizens can do about it.
In this article, we will try to explore three key aspects of disinformation in the digital age:
- how artificial intelligence and deepfakes create the illusion of reality;
- how social media algorithms determine what we see;
- how all of this is used as a tool of political influence and information warfare.
Because only understanding real threats can become the first step toward genuine information security.
1. Artificial Intelligence, Deepfakes, and Automated Disinformation: How a New Reality is Created
When we talk about disinformation in the 21st century, we are not just dealing with distorted facts. We face an entirely new phenomenon: content that never existed but appears real. These are generative artificial intelligence systems. Simply put, these are machines that not only process information but also create it: inventing images, writing news articles, imitating voices, and even generating faces that have never existed. And these very tools lie at the core of the new generation of disinformation campaigns.
AI That Creates Illusions
One of the most powerful technologies used to create fakes is Generative Adversarial Networks (GANs). These are behind the so-called deepfakes—a combination of the words “deep learning” and “fake”—videos in which famous people “say” things they never actually said. For example, in 2018, a video went viral in which U.S. President Barack Obama supposedly makes absurd statements. This video was created by researchers to demonstrate the threat of the technology, but just a few years later, such tools became publicly accessible. Today, a deepfake can be generated online in 10 minutes.
One of the most dangerous examples is the fake video of Ukrainian President Volodymyr Zelenskyy, in which he appears to call on Ukrainians to lay down their arms. The video was spread via Russian Telegram channels and social media in March 2022. Although the quality was poor, the very fact that it was created and disseminated during an actual moment of war proves that such tools have already become weapons.
At the same time, there are even more sophisticated versions: faces of non-existent people created using GANs are used to mask fake social media accounts. In 2020, The Daily Beast uncovered dozens of such profiles promoting pro-Russian narratives in the United States. The faces looked entirely plausible but had no real individuals behind them. The Daily Beast викрив десятки таких профілів, що просували проросійські наративи у США. Обличчя виглядали цілком правдоподібно, але не мали жодної реальної особи за собою.
Bots, Fake Factories, and Automated Propaganda
But content generation is only half the problem. Even more important is mass distribution. And here another technology comes into play—automation. Simple scripts, armies of bots, and fake accounts can spread disinformation through thousands of reposts in a matter of minutes. So-called “troll factories” (such as the infamous Internet Research Agency [IRA] in St. Petersburg, Russia) use artificial intelligence to automatically manage accounts, adapt messaging, and even interact with real people in discussions.
According to a study by the Stanford Internet Observatory, during the 2020 U.S. presidential election, activity was recorded from over 40 coordinated networks, some of which used automated platforms to simulate organic support for certain narratives. Some of these networks were linked to foreign governments, while others were associated with domestic radical groups.
Interestingly, modern bots no longer simply spread pre-written messages. They interact—responding to criticism, mimicking emotions, and creating discussions. This is no longer propaganda in the style of the Soviet newspaper Pravda—it’s a simulation of real dialogue that gradually shifts the tone of the public space.
The Scale of the Problem
Technologies for generating text, images, and video are not a fringe threat. For example, according to estimates by the research group Graphika, in 2023 the number of fake accounts created using AI exceeded 100,000 per month on just one platform. And that’s only the detected portion. Today, we live in 2025—and the scale of the problem is significantly larger!
But that’s still not the whole issue, because today the tools for creating deepfakes are becoming increasingly accessible. Just five years ago, it required technical knowledge—now practically any school student can use a free app to make a parody (or offensive) video featuring a teacher, politician, or classmate they dislike. And while this may seem like a joke, in the context of mass politics or warfare—it is far from humor; it is a weapon.
2. Social Media Algorithms: When Popularity Matters More Than Truth
Social networks have long ceased to be mere “showcases” for content. They have become the main “editors” of our informational reality. What we see in our Facebook, TikTok, X (Twitter), or Instagram feeds is not the result of chronology—it is the result of algorithmic selection. And therefore, a field of influence. This raises a key question: by what principles does this selection operate?
Emotion, Not Truth: The Main Criterion of Successful Content
Social media algorithms are designed with one goal: to keep users on the platform for as long as possible. This means that content that evokes emotion—anger, outrage, fear, excitement—is prioritized. These are the kinds of reactions that make people engage, leave comments, share posts—and thus increase views and time spent on the app.
In fact, back in 2021, former Facebook employee Frances Haugen exposed the company’s internal research, which showed that content that provokes polarization and negativity receives higher engagement. And although managers were aware of these effects, the platform did not change its recommendation principles—because it was not financially advantageous.
As a result, we face a situation where the algorithm does not distinguish between fake and real—it only recognizes engagement. If a fake news story about “secret laboratories in Ukraine” gets thousands of comments—it will be pushed further. If a calm explanation from an expert only gets a few likes—it disappears from the feed.
Algorithms as Catalysts of Disinformation
A 2018 study by MIT Media Lab showed that fake news spreads on Twitter six times faster than truthful information. The main reason: the emotional nature of fake content and its novelty. Humans instinctively respond to sensationalism. The algorithm simply amplifies this.
Particularly dangerous is the “explosive growth” principle—when an algorithm tests audience reaction to a new post in the first few minutes. If the reaction is active, the post gets wider reach. In this way, a single disinformation post can receive millions of views within an hour—before fact-checkers have a chance to respond.
Platforms often claim they are fighting fakes. But in practice, their verification systems work slowly, while the algorithmic promotion of fake content happens lightning fast. A 2022 report by the Mozilla Foundation revealed that even after a video was flagged as fake, YouTube continued recommending it—the algorithm had not been changed.
Scale and Targeting: Precision Strikes on Vulnerable Audiences
One of the most alarming characteristics of modern social networks is their ability for microtargeting. This means that fake content can be directed at very specific groups—such as veterans, the elderly, students, or ethnic minorities. All of this is possible due to the collection of personal data, which allows algorithms to “know” us better than we know ourselves.
In the U.S. and Europe, this has already had political consequences. For example, the Brexit campaign in the United Kingdom and the 2016 U.S. presidential election became examples of how Facebook was used for pinpointed disinformation campaigns, including through the involvement of Cambridge Analytica. And although that scandal drew the world’s attention, the mechanisms used at the time remain relevant to this day.
The Illusion of Choice: When We Don’t Choose What We See
Users often believe that they see in their feed exactly what interests them. But this is an algorithmic illusion of choice. What seems to be “organic” or a matter of “personal preference” is actually the result of millions of calculations that bring to the top not what is useful, but what is marketable. And unfortunately, disinformation today “sells” very well.
This leads to yet another problem—the so-called “information bubbles.” Algorithms create an environment around us where we only see content that confirms our views. This reduces critical thinking and makes us more vulnerable to manipulation.
The TikTok Case: A Perfect Platform for Viral Manipulation
TikTok is one of the most influential platforms among young people, particularly in the United States. The For You Page (FYP) recommendation algorithm has an incredible ability to make any piece of content go viral in an instant. This is both its strength—and its weakness. Back in 2022, NewsGuard discovered that when users entered basic queries related to the war in Ukraine or vaccinations, TikTok offered fake content among the top search results.
A unique feature of TikTok is that its algorithm can promote videos even if the creator has no followers—if a video captures the attention of just a few hundred viewers, it can quickly spread to millions. This creates an ideal environment for manipulative clips, deepfakes, or short fake explanations presented in an entertaining format.
In 2023, Forbes published an investigation proving that TikTok covertly moderates political topics in the U.S.—for example, it can “downplay” subjects related to Taiwan, the Uyghur genocide, or amplify anti-Western narratives about the war in Ukraine. Although the company claims neutrality, its ties to parent company ByteDance and the Chinese government have raised concerns in the U.S. government.
Meta: A Business Model That Encourages Fakes
Despite the scandals surrounding disinformation on Facebook, changes have been limited and selective. Although Meta has created fact-checking platforms and claims to be fighting fakes, its algorithmic priorities remain unchanged: content that evokes strong emotions spreads faster.
In 2023, the Center for Countering Digital Hate analyzed the top 100 most-engaged posts about the war in Ukraine—30% of them contained disinformation or pro-Russian messages, and most were videos or memes shared via Instagram Reels or Facebook Watch. The platform did not remove them, citing a “lack of policy violations.”
Even when fact-checkers flag a post as false, users can bypass restrictions—by posting screenshots or rephrasing the content. The algorithm cannot independently recognize distorted information, and therefore, fake content continues to “live on.”
X (formerly Twitter): Chaos After the Dismantling of Moderation
After Elon Musk acquired Twitter, the platform lost a significant portion of its moderation team. In the name of “free speech,” accounts previously banned for disinformation or hate speech were allowed to return. Within a few months, the number of disinformation campaigns on X increased sharply.
In 2023, the European Commission officially stated that X was the platform with “the highest level of disinformation activity among all social media in Europe.” Research showed that accounts actively spreading pro-Kremlin messages received increased reach, as the platform promoted subscriptions to Blue Verified—opening the door for mass disinformation to spread through seemingly legitimate accounts.
These examples illustrate that each social network has its own algorithmic peculiarities, but they all share one thing in common: engagement is prioritized over accuracy. And that is the root of a profound systemic problem.
YouTube Shorts: When a Video Fake Fits in 60 Seconds
In the context of the war in Ukraine, YouTube Shorts has become a powerful tool for manipulation. The short video format is ideal for edited clips, removing phrases from context, using deepfakes or audio recordings that are difficult to verify instantly. The YouTube algorithm aggressively promotes videos with high retention rates—regardless of their accuracy.
In particular, in 2023, the OSINT research group Bellingcat documented dozens of cases where pro-Russian channels published fake footage of shelling, compiled from the video game Arma 3 or old videos from Syria, presenting them as “new Ukrainian strikes.” These videos were immediately featured in recommendations and gained hundreds of thousands of views within hours. YouTube’s automated moderation system could not react in time.
Telegram: A Toxic Symbiosis of News, Emotion, and Anonymity
Telegram, especially under wartime conditions, has transformed into a central platform for real-time (and often fake) information. Anonymous channels such as Rybar, Voenkor Kotenok, or Rezident UA have audiences in the millions and distribute both actual news and targeted disinformation. Telegram’s algorithm officially does not filter content, and the forward-based format allows information attacks to be launched instantly.
Telegram has proven particularly effective in spreading panic—for example, during attacks on Kyiv or Kharkiv, when messages were circulated about “leaked intelligence,” “betrayal,” or “massive breakthroughs” by Russian forces. These were often coordinated information operations using bot farms to create the illusion of widespread panic.
Despite efforts by Ukrainian government bodies to run official channels and quickly debunk fakes, Telegram remains an environment with a high level of informational noise—and minimal restrictions when it comes to information warfare.
3. How States, Intelligence Agencies, Political Campaigns, and Lobbying Structures Organize Disinformation Through AI, Social Media, and Algorithms
In the digital age, information has become a tool of geopolitics—just as crucial as energy or military power. From state actors to private political campaigns, modern players have learned to systematically influence public opinion through social media platforms, using the capabilities of artificial intelligence and algorithmic systems for mass, targeted, and often hard-to-detect interference in public discourse, elections, and civil conflicts.
3.1. Intelligence Services as Operators of Information Warfare
Modern authoritarian states—primarily Russia, China, and Iran—have long and actively used disinformation as part of hybrid warfare. This includes:
- the creation of thousands of fake accounts using facial image generators (e.g., ThisPersonDoesNotExist); ThisPersonDoesNotExist);
- the use of LLMs (large language models like GPT) to write convincing comments, analytical pieces, and fake articles;
- publishing materials on proxy websites that pose as independent media outlets;
- launching coordinated networks of bots and trolls via Telegram, Facebook, YouTube, TikTok, and X (Twitter).
For example, Russia has focused on aggressive information tactics. Well-documented operations include: - interference in the 2016 and 2020 U.S. elections via the Internet Research Agency (IRA) and affiliated entities;
- the fake “Zelenskyy surrender” deepfake video in 2022;
- campaigns to discredit Ukrainian refugees in the EU;
- promotion of fakes about “U.S. biolabs in Ukraine,” “crucified children,” “destruction of Christians,” and similar narratives.
China натомість діє більш обережно – робить ставку на довгострокове формування позитивного іміджу КПК і поширення антагоністичного до США порядку денного через контрольовані медіа (CGTN, Global Times), дипфейкові відео та боти в TikTok, Weibo та навіть на Reddit і YouTube.
Iran employs similar tactics, promoting anti-Western messages through a network of proxy websites and Facebook pages targeting leftist or religious audiences in the United States.
3.2. Political Campaigns and Super PACs: Tools of Legal Manipulation
In the United States, domestic political actors also use methods that, in practice, border on disinformation—though formally they are considered “political marketing.”
Super PACs—independent political action committees—have become particularly dangerous. They are allowed to spend unlimited amounts on campaigns without directly coordinating with candidates. They commission:
- neural network-generated negative advertising;
- targeting based on fears, biases, or ethnic/cultural identity;
- selectively edited videos to discredit opponents (often using deepfakes or out-of-context quotes);
- the use of bots and “community pages” that mimic grassroots movements.
In 2020, several states uncovered disinformation campaigns about mail-in voting. Although these campaigns were formally unaffiliated with any political party, they effectively supported the strategic interests of one of the two major political forces.
3.3. Lobbying and Ideological Groups: A Slow but Systemic Erosion of Information Hygiene
Another channel for the spread of disinformation includes think tanks, religious groups, and radical media outlets that, under the guise of offering “alternative viewpoints,” disseminate distorted or manipulative information. In the U.S., such entities often include:
- certain media outlets that fail to verify or intentionally manipulate facts;
- influencers with large followings who spread information without fact-checking;
- YouTube channels and Telegram groups that exploit themes such as the “deep state,” “vaccination,” or “the media as the enemy of the people.”
These channels often exploit algorithmic loopholes: dramatic headlines, emotionally charged videos, anger, fear, or a sense of “us versus everyone”—all of which work effectively within social media recommendation systems.
3.4. Information Attack as a Business Model
In today’s world, disinformation is not only a political tool—it is also a profitable business. Content that provokes outrage, fear, or hatred generates more views, more likes, and more revenue through monetization. This creates a unique alliance between those who seek to manipulate and those who simply want to make money.
Even without direct interference from intelligence agencies, the system perpetuates itself—algorithms amplify extremes, audiences become radicalized, and platforms profit.
4. Political Perspective: Information as a Tool of Power and Hybrid Warfare
In the 21st century, information is not just knowledge—it is also a tool of pressure, manipulation, and control. Authoritarian states, intelligence services, populist campaigns, and even private political entities actively exploit the opportunities of the digital environment to shape agendas that serve their interests.
Digital platforms are no longer merely arenas for discussion—they have become battlefields, where content, audiences, and algorithms are transformed into strategic assets. For the first time in history, a single computer in some authoritarian country can massively influence the political situation in Washington or Brussels.
Russia: An Industrial Disinformation Machine
Russia has been systematically developing its infrastructure for information operations for over a decade. From troll factories (like the notorious Internet Research Agency, IRA) to modern AI-generated accounts, the Kremlin views the information space as a theater of war. For example, in a 2020 report by the U.S. Senate Intelligence Committee, it was established that Russian entities actively interfered in the 2016 and 2020 U.S. elections through Facebook, YouTube, Instagram, and Twitter, using both fake profiles and targeted ads.
In 2022–2023, according to EU DisinfoLab, Russia used AI to create thousands of fake accounts, generating realistic-looking faces with neural networks (thispersondoesnotexist.com) and writing texts using LLMs (large language models). Bot networks published fabricated stories about U.S. biolabs in Ukraine, “NATO conspiracies,” or “massive Ukrainian military losses,” which were then picked up by “media proxies”—anonymous Telegram channels or junk websites.
Particularly dangerous is the use of deepfakes, including videos in which “Zelenskyy” allegedly surrenders or “Biden” makes false statements. Although these videos are still technically imperfect, their strength lies in the immediate effect they have on an audience unprepared for critical evaluation.
China: Control, Censorship, and Narrative Export
China’s strategy is more systematic: the state tightly controls the internal information environment (via the “Great Firewall”) while simultaneously exporting propaganda abroad. Platforms such as CGTN, Global Times, and People’s Daily conduct English-language campaigns, often disguising propaganda as “alternative viewpoints.”
In 2023, research by the Brookings Institution documented how pro-Chinese accounts used AI to promote “soft power,” particularly in regions of Africa, Asia, and Latin America. For example, propaganda clips about the “successes of Chinese medicine,” the “failure of democracy in the U.S.,” or China’s “non-colonial foreign policy” were created by AI in dozens of languages and promoted through platforms such as TikTok and Weibo.
Although China is not as aggressive in targeting U.S. elections as Russia, it is gradually competing in the battle for global narrative dominance—using the very same technologies.
It is important to understand: algorithms do not inherently recognize what is “true.” They amplify what generates reaction—and therefore, any actor with the resources to create effective (even fake) content can win the war for attention.
5. The Response of Journalism, Social Media, and the Public to Disinformation: Challenges of the Media Ecosystem
In a situation where hybrid disinformation spreads at record speed, the role of journalism, independent media, digital platforms, and an active civil society becomes existential. This is not just a fight for the truth—it is a fight for a shared reality, without which democracy cannot function.
5.1. Professional Journalism Under Pressure: Speed vs. Verification
In the digital age, traditional journalism faces an unprecedented challenge: the speed of information devalues the process of verification. In the race for clicks, even major newsrooms sometimes publish unverified materials, relying on viral videos or social media “news.” This is particularly dangerous in an era when AI can generate entirely realistic—but fake—content in a matter of seconds.
This intensifies the debate over ethical standards and editorial protocols: how to verify AI-generated content, whether to label images that may have been created by neural networks, and how to fact-check videos on platforms like TikTok or Telegram.
5.2. The Role of Fact-Checking: Important, But Too Late?
Fact-checking platforms like Snopes, PolitiFact, Bellingcat, StopFake, and The Markup have become critically important players. Their investigations help to stem the tide of fake news, especially regarding:
- elections;
- topics related to vaccination;
- disinformation about Ukraine;
- distortions of context in international events.
However, there is a structural problem: verification takes hours or even days, while fake news spreads within minutes—and often irreversibly. By the time a claim is debunked, a person may have already formed an opinion. This is the so-called first impression bias.
For example, back in 2022, a Facebook page with 50,000 followers shared a deepfake video in which General Zaluzhnyi allegedly admits defeat in Bakhmut. The post received 15,000 shares before it was debunked by StopFake and taken down.
5.3. Platform Responsibility: Algorithms That Fuel Radicalization
The algorithmic logic of social media platforms—TikTok, YouTube, X (Twitter), Facebook—is designed to maximize user retention. These systems do not distinguish between truth and lies—they amplify whatever provokes a reaction:
- YouTube and Facebook have acknowledged that conspiracy theories and radical videos have longer average watch times and are therefore promoted automatically;
- TikTok in 2023–2024 massively spread anti-Ukrainian videos that were later revealed to be edited or AI-generated;
- Telegram has become the primary channel for unfiltered information in the post-Soviet space, and its algorithms are almost entirely unregulated.
The response has included: - labeling AI-generated images and videos (initiatives by Meta and Google in 2024);
- sets of rules for content transparency—particularly the EU Digital Services Act;
- partnerships with fact-checkers, such as Meta’s collaboration with Instagram.
However, these steps are often significantly delayed or only partially effective.
5.4. Civic Initiatives and Digital Education
Hundreds of grassroots initiatives (from the English grassroots—literally “roots of the grass,” meaning bottom-up or community-level efforts) are emerging to promote critical thinking, including:
- school-based media literacy programs;
- public campaigns like “Don’t Share Before You Check”;
- volunteer networks of digital activists (such as InfoDefenders or Cyber Volunteers for Ukraine);
- apps that automatically verify links, like NewsGuard or SurfSafe.
Ukrainian media education projects in the U.S. diaspora are also active—offering webinars, lectures, and TikTok influencers who explain how to recognize AI-generated fakes, how to use fact-checking tools, or how to avoid falling for manipulations on Telegram channels.
5.5. Legislative Response: Delayed and Fragmented
Many democratic countries are attempting to regulate AI and disinformation at the legislative level. The European Union has already introduced:
- the Digital Services Act (DSA)—requiring algorithmic transparency, effective content monitoring, and rapid response to disinformation;
- the AI Act—which imposes strict limitations on generative AI in high-risk areas, such as politics, education, and elections.
In the U.S., serious legislative discussions only began in 2023–2024 about laws that would obligate tech companies to:
- label AI-generated content (e.g., video or audio files created using AI);
- protect elections from manipulation through deepfakes;
- prohibit the use of AI to mislead voters.
However, strong lobbying from Big Tech (Meta, Google, X, TikTok, and others) significantly slows these efforts. As a result:
- the U.S. still has no nationwide law explicitly banning the use of political deepfakes during election campaigns;
- platforms remain minimally accountable for disinformation—mainly due to Section 230 of the Communications Decency Act (passed in 1996), which grants digital platforms immunity from legal liability for user-generated content.
In today’s world, where artificial intelligence, deepfakes, and social media algorithms are radically transforming the media landscape, disinformation has become one of the most serious challenges to democracy and security. Technological innovations that unlock new possibilities also bring with them new dangers—from the rapid spread of fake content to large-scale manipulation of public opinion.
An analysis of the role of AI and social media in this process shows that disinformation is not merely an accidental error or technical glitch. It is a deliberate hybrid warfare tactic, in which states, intelligence services, political campaigns, and lobbying entities structure and scale fake narratives using algorithms and the capabilities of artificial intelligence.
At the same time, the media system and civil society are not powerless. Journalism, fact-checking, educational initiatives, and legislative efforts represent a front where the battle for truth is being waged. However, many of these measures remain insufficient or react too slowly—demanding new, more proactive strategies.
The key takeaway is that the fight against disinformation is a shared responsibility—the responsibility of every individual citizen, every democratic institution, and every tech platform. The active stance of the Ukrainian diaspora in the United States is especially important, as it can serve as a model of both awareness and effective resistance to information attacks.
Only through continuous improvements in media literacy, legislative regulation, and international cooperation can we minimize the destructive impact of fake news, preserve democratic values, and ensure information security on a global scale.
About Author:
