How Foreign Governments Use Social Media to Shift Public Opinion
The idea of foreign governments trying to influence public opinion in other countries is not new. State-sponsored radio broadcasts, pamphlets, and propaganda films have been around for a century. What is new is the scale, the precision, and the invisibility of how this gets done today. Modern influence operations do not look like propaganda. They look like your neighbor’s Facebook post.
The basic playbook has been documented extensively since the 2016 US election exposed it. A foreign government, or a contractor working on its behalf, creates a network of social media accounts. Some of these accounts pose as ordinary citizens of the target country. They post content designed to appeal to specific segments of the population, often content that exploits existing social divisions: immigration, race, economic anxiety, political identity. The goal is not necessarily to promote a specific policy, but to increase division, lower trust in institutions, and make democratic discourse more chaotic and less functional.
The accounts that do this most effectively do not lead with the divisive content. They spend months building authentic-looking profiles: posting about local sports teams, sharing recipes, engaging in normal social activity. Once the account has followers and a track record, the divisive content gets introduced gradually. By the time the influence operation is in full swing, the accounts look indistinguishable from the organic community they have been cultivating.
Several countries have developed sophisticated versions of this capability. Russia’s Internet Research Agency, which was indicted by a US grand jury in 2018, operated accounts that collectively reached tens of millions of Americans before the 2016 election. China runs influence operations targeting diaspora communities and shaping narratives around Taiwan, Hong Kong, and Xinjiang. Iran has run operations targeting both US and Israeli audiences. The tactics differ in emphasis, but the structure is similar: scale, patience, and plausible deniability.
What makes this particularly hard to counter is that the content itself is often not false. Divisive things are real. Grievances are real. An influence operation does not need to invent tensions from scratch. It just needs to find existing tensions, find the people who feel them most strongly, and amplify. A real story about a real injustice can be weaponized for influence purposes just as effectively as a fabricated one, and it is harder to fact-check your way out of.
PaxPoint gets into the genuine uncertainty about effectiveness. Not all operations achieve their goals. Not all amplification translates to changed minds. But the documented scale of the attempts justifies taking the phenomenon seriously even before the outcome data is fully clear.
The practical question for ordinary social media users is how to engage with this knowledge without becoming so paranoid that you distrust everything. A few useful habits: pay attention to whether an account has a history before the content you are seeing, be more skeptical of content that seems designed to provoke a strong emotional reaction about a divisive topic, and consider whether what you are reading is trying to inform you or trying to make you angry at a specific group.
Social media companies have made some investments in detecting and removing coordinated inauthentic behavior, with mixed results. The arms race between detection and evasion is ongoing. The most reliable protection is a combination of platform effort and individual awareness, and right now the individual awareness side has a lot of room to grow.