It is hard to imagine President Donald Trump coming to power without social media. His political genius has been to master the medium of our age, stoking grievances, mobilising support and provoking outrage in unrivalled ways.
During his first term, he tweeted more than 25,000 times (averaging 18 posts a day). He has been just as finger happy on his own social media site Truth Social in his second term. One could almost say his approach is: government of social media, by social media, for social media.
Trump’s Maga movement has also found its natural home on social media, with many thousands of accounts amplifying his messages. That makes it all the more jarring to discover that some of the most active “America First” accounts are run from abroad.
In a move to secure “the integrity of the global town square”, the social media platform X last Friday began posting user location data. As a result, it emerged that dozens of influential Maga accounts are run out of foreign countries, including Russia, India and Nigeria.
For example, the MAGA NATION account, which claims to be a “Patriot Voice for We The People” with more than 393,000 followers, is based in eastern Europe (Non-EU), X revealed.
Malign foreign actors are known to use imposter accounts — either to manipulate political debate or to generate traffic and make money. That is just one of the ways in which our infosphere is being deliberately degraded.
There are three other types of social media deformities, too. Call them the four horsemen of the infocalypse. Unchecked, they will surely destroy our trust in almost anything we read online.
The second corrosive influence is how extremist views, once confined to the darker corners of the web, have seeped into mainstream debate, as documented by Julia Ebner, a researcher at Oxford university and author of Going Mainstream.
In a speech in London this week, Ebner described how Islamist jihadis, misogynistic incels, conspiracy theorists and white supremacists followed a similar digital playbook to spread their views. Extremists would often find each other on sites such as 4chan, Parler and Discord, where they attract followers through gamification, globalisation and glorification before migrating to mainstream platforms such as Facebook, X and YouTube, to popularise their views.
Often, their ideas would be fuelled by media sites and troll armies from Russia, China and Iran, Ebner said. She pointed to information on the Russian sites Sputnik and RT, which she said supported vaccination drives in Russia but fanned anti-science conspiracy theories in English and German.
The influence of extremists is now being magnified by generative artificial intelligence tools, which have cut the cost of producing content close to zero. This has resulted in the third influence — a vast diffusion of AI-generated slop. Two years after the launch of ChatGPT in November 2022, the number of AI-generated articles published on the web surpassed human-written articles.
According to the AI agency Graphite, at least 52 per cent of internet content was machine-generated by May this year. Not only does this AI-generated content threaten the integrity of the web, it could also endanger the resilience of future AI models.
If AI models ingest too much synthetic data then there is a danger they might go “mad” — or suffer from model autophagy disorder. When the input for new models increasingly becomes the output of older models then it risks model collapse, says Ethan Smith, Graphite’s chief executive.
The fourth way in which trust is being corroded is when social media platforms themselves act more as partisan publishers than neutral platforms.
Shortly after Elon Musk acquired X (then Twitter), he scrapped most of its content moderation machinery, leading to a surge in racist and divisive speech, according to New York Times analysis.
X’s latest move to post data about accounts suggests it is more serious in exposing “inauthentic” users. Then again, that has been standard operating procedure within the safety teams of other social media services for years.
The prospect of AI-enabled extremists and hostile foreign powers further exploiting the vulnerabilities of permissive social media companies is disturbing. But while she is a short-term pessimist about these trends, Ebner declares herself a long-term optimist.
There is now a greater recognition of these problems among governments and regulators. Innovative platforms are also emerging that can operate as conversation networks, encouraging collaboration more than confrontation. AI companies and social media sites are attaching a higher value to human-generated content. And there is an increasingly healthy scepticism of what users see online, especially among the young.
“I have a lot of hope that the next generation will show more and more resilience. They are intuitively familiar with how these technologies work,” Ebner tells me.
Moreover, other politicians are also mastering social media. New York’s newly elected mayor, the 34-year-old democratic socialist Zohran Mamdani, mobilised massive support through viral videos. “Winning New York now means speaking fluent TikTok,” wrote one observer.
No matter what the flaws of social media, it is a lesson no politician can afford to ignore.