In April and May, hundreds of users of the social media platform Bluesky discovered that their accounts had been hijacked and used to spread content about Ukraine. Researchers at Clemson University later attributed the campaign to the Social Design Agency, a Russian consulting firm sanctioned by the U.S. Department of Treasury for running Kremlin-directed malign influence campaigns. The Bluesky effort turned out to be only one piece of the SDA’s much larger portfolio of recent operations: leaked documents obtained by the news portal Delfi Estonia and shared with the Organized Crime and Corruption Reporting Project revealed plans for real-world acts of vandalism, intimidation, and election interference.
Most discussions of online influence and disinformation campaigns assume that attacks such as the one on Bluesky are discrete efforts. But as the SDA’s operations show, influence operations are increasingly interconnected and part of propaganda strategies that transcend traditional boundaries. Sophisticated states such as China, Iran, and Russia routinely combine overt public messaging with covert influence and mix broadcast channels with social media platforms. State media outlets, official government accounts, paid influencers, fake personas, hacked accounts, and bot armies are parts of the same system. This integrated approach, which we call “full-spectrum propaganda,” is now the norm for U.S. adversaries trying to spread their message abroad.
In full-spectrum propaganda, as we outlined in a recent Security Studies article, the many channels that states can use reinforce each other, making narratives seem more credible and allowing them to proliferate more widely. A story publicized on a state-backed broadcast channel may reach one audience, while social media accounts catch the attention of another. Official outlets can quote from posts on fake accounts whose creators masquerade as ordinary members of the public, making a fringe claim seem normal. Paid influencers can tailor messages in a style that resonates with niche audiences. The rapid expansion of artificial intelligence is making this strategy even more powerful: states are increasingly integrating AI into every level of their propaganda operations, allowing small groups or individuals to conduct propaganda campaigns that would previously have required far greater resources.
Strategies that merely address discrete foreign influence campaigns will fail to meet the challenges presented by full-spectrum propaganda. The United States needs to be able to understand, assess, and counter influence operations holistically, yet as the threat grows, the United States has been undercutting the capacity that would allow it to do so. Politicians have been dismantling the government agencies designed to track foreign influence, and social media companies are raising barriers to research on their platforms, leaving the country unprepared. As AI accelerates the spread and expands the reach of global propaganda, Washington is blinding itself to the challenge.
EVERYTHING EVERYWHERE ALL AT ONCE
For much of the last decade, politicians and the public understood foreign influence operations on social media through the parts that first became visible. Most often, this would be a network of fake accounts discovered by social media companies as part of their efforts to eliminate bots, scams, and online deception on their platforms. After taking down such accounts, companies would write reports and researchers (including us) would publish analyses to explain the evolving tactics, techniques, and procedures of online influence campaigns. The focus on the specifics of each individual operation led people to think of the threats as narrow, distinct incidents.
But the fragment of what is visible is not the whole story. State-backed propagandists do not have to choose between state media and fake accounts, between covert personas and overt broadcasters, or between a spoofed media outlet or a Telegram channel. They can use all of them together—and they are increasingly doing so.
The SDA, for instance, documented its own work in the style of a consultancy reporting to a client. That client, the leaked documents suggest, was the administration of Russian President Vladimir Putin, whose officials appear in the agency’s internal communications overseeing its work. The consulting firm’s covert operations are just one component of a much broader Russian propaganda apparatus that also includes state media networks, official government channels, and fake personas. Moscow and its proxies use all of these tools in tandem to amplify propaganda messages and reach many types of audiences.
Looking at foreign influence campaigns as a social media problem is too limited.
As the Organized Crime and Corruption Reporting Project and other news outlets explained, the SDA was behind a fake video in June 2025, branded to appear as if it were made by the Saudi media outlet Al Arabiya, which claimed that Ukrainian President Volodymyr Zelensky purchased a luxury apartment in Burj Khalifa, a skyscraper in Dubai. An internal report estimated that this false story could have logged as many as 86 million views globally; a similar SDA effort targeting Armenian Prime Minister Nikol Pashinyan likely racked up over ten million views. In some cases, rather than mimicking an existing outlet such as Al Arabiya, SDA conjures fully branded fake news outlets: one leaked document described plans to use an outlet tied to the SDA in Armenia’s June parliamentary elections.
When more distance is required, SDA turns to paid influencers. Earlier this year, for example, the Kremlin wanted to support the reelection bid of Viktor Orban, then Hungary’s prime minister. According to the Financial Times, the SDA, recognizing that overt support from Russia could be interpreted as foreign meddling and hurt Orban’s campaign, proposed contacting local influencers via go-betweens to make the Russian content appear homegrown.
The SDA even bridges the gap between the real world and online information by staging controversial or threatening events to generate attention. In September 2025, severed pig heads, including several marked “Macron” for the French president, were left outside mosques and Islamic centers in and around Paris. The leaked SDA files reveal the meticulous plans behind the effort, along with similar vandalism plots elsewhere in France and in Germany. Mainstream broadcast and social media covered these events, generating real-world concern about xenophobia and Islamophobia in Europe. The SDA calls these “cognitive strikes”: the point is securing visibility, shaping perceptions, and creating emotional responses that can be exploited.
Such covert activity complements Moscow’s efforts to continue to develop its broadcast propaganda apparatus worldwide. In 2025, the Kremlin allocated 31 billion rubles ($367 million) to RT, the Russian state broadcaster. RT and Sputnik, the state news agency and radio service, have expanded across Africa, Latin America, the Middle East, and Southeast Asia, according to the BBC, including by taking over local media channels. RT Academy, the broadcaster’s educational arm, ran a journalism program in Jakarta, the Indonesian capital, to train a new generation of reporters; its programming included “fact checking” and “applying ethical standards” in storytelling. RT is not only broadcasting on its own stations but is also funding and operating sympathetic proxy channels. In 2024, for example, the U.S. Department of Justice indicted two RT employees for funneling nearly $10 million to Tenet Media, a right-wing U.S. media outlet. Even influencers working for Tenet Media were unaware of the source of that funding.
A CONFLUENCE OF INFLUENCE
Full-spectrum propaganda presents a distinct challenge because it exploits the way modern information systems confer legitimacy. As research in psychology has shown, repetition reinforces belief, and a narrative may seem more believable when it appears in multiple distinct places or when many seemingly unconnected people start talking about it. A claim looks more credible when an outlet, a few influencers, and some individuals in a Telegram channel seem to arrive at the same conclusion independently.
Integrated propaganda campaigns can also more easily evade existing crackdown efforts by individual platforms. Because social media companies differ in how they regulate content, when fake accounts on one site get removed, they may simply respawn or migrate to platforms with more permissible policies or weaker enforcement capacity. When Russian officials and state media outlets portrayed the bombing of a Ukrainian maternity hospital in March 2022 as a faked event and claimed that the victims were paid actors, social media platforms such as Twitter took down the deceptive content. But anonymous Telegram channels claiming to be fact-checking organizations gained popularity and spread the false narrative there. Some of these efforts may have been genuine attempts to investigate the charges, but many others appeared to be the work of state-backed propaganda groups seeking new places to share their message.
In a world of full-spectrum propaganda, then, looking at foreign influence campaigns on a platform-by-platform basis or simply as a social media problem is too limited. Merely analyzing posts on Bluesky, for instance, overlooks the way the message they are promoting connects to simultaneous efforts on Facebook and X, articles in state media, and public broadcasts around the world. Understanding a modern propaganda campaign requires examining where else a narrative may have appeared, whether fake outlets were quoted by influencers, whether ostensibly independent sources were connected, and how those different parts of the system reinforce one another.
AI is further transforming propaganda strategies and making the challenge of countering it harder. An investigation by The New York Times revealed that China and Iran have used Chinese open-source models to run propaganda campaigns through AI agents with limited human supervision. The agents created fake accounts on a variety of social media networks, posted false stories, and coordinated their messaging. Although these examples were limited in reach, they are likely harbingers of what is to come. State media channels are also using AI to expand their reach: Anthropic found that the Russian state media had used its AI model, Claude, to turn Romanian and Moldovan news into Russian-language articles, showing how states can easily translate content without needing linguistic expertise. Propagandists are even targeting the AI systems themselves to try to shape the answers that chatbots provide. They are creating fake websites and publishing content designed to come to the surface when AI models retrieve information and to influence the data that trains AI models on how to respond.
CAPACITY COLLAPSE
Because U.S. adversaries are pursuing a full-spectrum, AI-optimized approach to propaganda, Washington needs as much capacity as it can get to assess the threat holistically and expose or combat it. Unfortunately, social media platforms and the U.S. government have spent the past several years dismantling exactly that capacity, leaving the United States unaware of its vulnerabilities.
In Silicon Valley, companies have made it harder to access data about their platforms, limiting the visibility of propaganda campaigns. Social media platforms have walked back platform integrity and transparency commitments and dissolved collaborations with researchers. In May, Meta laid off ten percent of its workforce, including members of its investigations team, which had been instrumental in combating online influence operations. In 2024, Meta eliminated CrowdTangle, a transparency tool that researchers could use to gain access to public posts, and in 2023, Twitter, now known as X, ended free access to its API for researchers, who had used it to study public posts on the platform. X now charges them $5 for access to every 1,000 tweets, or at least $42,000 per month for an enterprise subscription, making it far too costly for many academics to conduct research on the platform at scale.
These barriers have limited independent checks on statements from the platforms and the government about foreign influence, making it harder to identify the severity of such campaigns and draw useful lessons from them. When in August, X announced the removal of 200,000 accounts from a suspected Chinese bot farm—200 of which it claimed posted information about AI data centers—the company did not make data about the operation available to researchers. Politicians seized on the news to blame Chinese covert influence for the backlash to data centers in the United States, but an analysis conducted by Clemson University professor Darren Linvill revealed that posts from the accounts that could be identified had received virtually no engagement and were not a source of significant influence.
Influence operations will increasingly involve AI-enhanced systems.
In Washington, a focus on fighting domestic censorship has led to the dissolution of institutions leading U.S. efforts to track foreign influence campaigns. U.S. President Donald Trump’s executive order on ending censorship, signed on his first day in office in January 2025, was framed as an attempt to prevent the government from infringing on constitutionally protected speech. But the result has been to stifle federal efforts to mitigate foreign influence, and claims of censorship have been used to shut down or weaken essential institutions. In 2024, Congress failed to extend the mandate for the State Department’s Global Engagement Center, an office set up in 2016 with bipartisan support and a goal of tracking propaganda from China, Iran, Russia, and other foreign actors. The organization created to replace it, the Counter Foreign Information Manipulation and Interference hub, was then shut down by the State Department. Republican lawmakers and other critics who campaigned against these offices claimed that they were trying to censor Americans, but no courts ultimately found that they had.
Then, in August 2025, Director of National Intelligence Tulsi Gabbard closed the Foreign Malign Influence Center, which coordinated the intelligence community’s analysis of foreign interference and methods for countering it. The Justice Department also shut down the FBI’s Foreign Influence Task Force, which sought to expose illegal foreign interference on U.S. soil and aligned investigations and intelligence sharing within and across agencies. Taken together, the destruction of these institutions has created a huge gap in the knowledge available to people in Washington about what China, Iran, and Russia are doing, especially the entire spectrum of influence activities they are undertaking. As Senator Mark Warner, vice chairman of the Select Committee on Intelligence, warned this spring, too little capacity remains to track or disrupt covert campaigns run by adversaries of the United States. At the same time that foreign influence operations are becoming more integrated, the U.S. government is losing the coordination needed to understand them.
As the United States blinds itself, foreign adversaries are expanding their efforts. Influence operations will increasingly involve AI-enhanced systems and organizations that can autonomously create fake news outlets, generate articles and videos, disseminate them through hijacked or inauthentic accounts, seed them among influencers, quote them in official state media, and trick AI answer engines into thinking that fallacies or outright lies represent the truth or mainstream opinion. As China, Iran, and Russia operate across the full spectrum of propaganda, the United States is getting stuck in narrow silos. Without the capacity to even understand the nature of the threat, Washington cannot begin to counter it.
Loading…
