Imagine you are a school principal in a quiet district in Peru. One morning, an official-looking email arrives, stamped with the seal of your country’s Ministry of Education. It announces a special national celebration, something called the National Day of Cultural and Linguistic Diversity, and asks your school to hold an assembly. Students are to learn about certain Ukrainian political figures, read prepared statements, and present them as part of the day’s activities. It sounds odd, but the document looks legitimate. It has the right headers, the right formatting, the right tone of bureaucratic authority. So you agree. Teachers print the materials. Students stand at the podium. And without anyone realizing it, a Russian influence operation has just used your classroom as a stage for geopolitical theater. This is not a hack. No servers were breached, no passwords stolen, no ransomware demanded. It is something far quieter and, in some ways, more dangerous: a carefully engineered deception designed to exploit trust in institutions, in official language, and in the everyday routines of public life. Dark Reading reports that this is exactly what has been happening in Latin America, as a Russian-linked operation has turned to artificial intelligence to impersonate educational authorities and sow political discord in Peru and Ecuador.
The operation, identified by OpenAI as Dark Clark, represents a new and unsettling blend of old-school influence tactics and modern generative AI. According to the report, the campaign used AI to draft reports, generate deceptive content, and produce fraudulent directives that were sent directly to school administrators. In Peru, the operation successfully convinced schools to hold a fake National Day of Cultural and Linguistic Diversity, an event that required students to highlight specific Ukrainian political figures in their presentations. The purpose was not educational. It was to create a plausible, localized event that could be filmed, photographed, and shared as evidence of Ukrainian influence in Latin America, or as proof of some hidden political agenda. In Ecuador, the same operation used similar tactics to orchestrate a false loyalty oath involving President Daniel Noboa and Erik Prince, a controversial American businessman. The oath was designed to look like an official school ceremony, complete with flags, pledges, and formal language. The goal was to make people believe that Ecuador’s government was aligning itself with foreign mercenary interests, and that the country’s youth were being indoctrinated. In both cases, the fake events were not just malicious content posted online; they were physical events, staged in real schools, involving real children and real teachers. That is what makes this so disturbing. It is one thing to spread a rumor on social media. It is another thing entirely to use trusted institutions to act out your disinformation in public.
What makes Dark Clark notable is not that it invented a brand-new kind of cyberattack. It did not. Instead, it used artificial intelligence as a force multiplier for familiar psychological operations. The operators claimed success in a number of regional influence efforts, including fabricated stories in Brazil and Bolivia, and they clearly understood the power of local context. They knew that a well-worded memo from a government ministry would be taken seriously. They knew that schools are often eager to participate in civic events and that administrators rarely question official-looking paperwork. They knew that teachers, busy and understaffed, would not have time to verify whether a National Day of Cultural and Linguistic Diversity actually existed. And they knew that once the event was held, it would generate authentic photographs and videos that could be circulated online. This is where AI became so valuable. It allowed the operators to produce linguistically fluent, culturally aware, and professionally formatted documents at scale. The AI did not need to invent a new attack vector; it simply made the old ones faster, cheaper, and harder to spot. As researchers noted, the technology primarily served to improve linguistic fluency and operational efficiency. In other words, AI did not change the game. It just gave the players better equipment. It is like a counterfeiter who still passes fake bills, but now has a better printer, better paper, and a sharper eye for detail. The underlying crime is the same, but it is far more convincing.
The ultimate goal of these activities, according to the report, appears to be damaging the international reputation of Ukraine. By planting stories in Latin America that tie Ukraine to questionable political figures, foreign mercenaries, or nationalist ideologies, the operation seeks to undermine the sympathy and support that Ukraine has received since Russia’s full-scale invasion began. This is not an accident. It is a deliberate strategy to expand the information war beyond Europe and into regions where public opinion is still forming. Latin America has become a battleground for influence, and schools are being used as unwitting accomplices. The choice of targets is telling. Peru and Ecuador are not major powers, but they are democracies with active civil societies and deep traditions of civic education. They are also countries where institutional trust, while fragile, still exists. By exploiting that trust, the operation hopes to create confusion, sow discord, and eventually shift public perception in ways that benefit Moscow. The operation is linked to entities associated with the late Yevgeniy Prigozhin, the Russian oligarch who funded the Wagner Group and who built an entire network of influence operations before his death. That connection matters. It shows that these tactics are not the work of amateurs or lone hackers. They are part of a coordinated, state-sponsored effort to weaponize information, and they are evolving.
Experts who study disinformation are careful to point out that AI is not magic. It does not create influence out of nothing. The underlying tactics remain rooted in traditional disinformation strategies: exploit trust, use official-looking documents, create confusion, and let the target audience do the work of spreading the message. What AI does is provide an incremental advantage in scale and editorial quality. It allows a small team to produce content that would have required a much larger operation a decade ago. It allows them to write in fluent Spanish, to adapt to local cultural norms, and to generate multiple versions of the same lie for different audiences. But the vulnerability that the operation exploits is not technical. It is human. People trust authority. People trust official language. People trust institutions. And when those institutions are busy, under-resourced, and unfamiliar with the tactics of foreign intelligence services, they become easy targets. The most effective defense, therefore, is not just better firewalls or stronger encryption. It is skepticism, verification, and media literacy. A school administrator who pauses to call the Ministry of Education and confirm the authenticity of a directive can stop an entire operation in its tracks. A teacher who asks why a civic event focuses on foreign political figures can raise the alarm. A parent who notices something odd and reports it can prevent the spread of fabricated evidence.
This story should be a wake-up call, not only for Latin America but for the entire world. It reminds us that cybersecurity is no longer just about protecting data. It is about protecting truth, trust, and the shared institutions that hold communities together. For security professionals, the essential knowledge is this: AI is not just a tool for defenders, but also for adversaries. It is being used to improve phishing emails, create deepfakes, generate fake news, and now, as we see here, to impersonate entire government ministries. The practical strategy is not to chase every new technology, but to focus on the fundamentals. Verify communications through independent channels. Establish clear protocols for how official directives are issued and approved. Train employees, educators, and administrators to recognize the signs of social engineering. Build partnerships with government agencies, civil society, and technology companies like OpenAI that are actively tracking these operations. And perhaps most importantly, understand that the target is not the machine. It is the person sitting at the desk, the principal reading the email, the teacher preparing the lesson, the student reciting the pledge. The best defense is a human one: a community that knows how to ask questions, verify sources, and resist the quiet erosion of trust. Dark Reading’s coverage of Dark Clark offers essential knowledge and practical strategies to use AI to better your security program, but the real lesson is deeper. In an age of synthetic media and algorithmic deception, the most valuable asset we have is not artificial intelligence. It is human judgment, exercised carefully, collaboratively, and with the courage to say, “This does not feel right.”