Social media giant Meta has published a report detailing observations made across its platforms following an unprecedented year of elections around the world which saw as many as two billion people expected to take to the polls.

Though many began the year fearful of the impact AI might have on elections, including around deepfakes and disinformation, Meta claims that these risks didn’t materialise in any significant way, and that any impact AI had was modest and limited on its platforms.

The company said that while there was some confirmed or suspected uses of AI in this way, the volumes typically remained low, and that existing policies were sufficient to tackle the threat.

In fact, Meta reported that during major elections, including the US Presidential race and UK General Election, AI content related to elections, politics, and social topics represented less than 1% of all fact-checked misinformation.

The firm also said that in an effort to prevent its Imagine AI image generator being misused to create election deepfakes, 590,000 requests to create images of leading US politicians were rejected.

Meta also claims that the use of genAI by so called ‘Coordinated Inauthentic Behavior’ networks, otherwise covert influence campaigns, was limited in scope, and did not result in major gains.

Meta’s findings back up earlier research from The Alan Turing Institute, which reported in September that it had found no evidence AI-enabled misinformation meaningfully impacted the results of the UK and European elections. 

However, following that research the Institute went on to urge policy makers, regulators, and social media platforms to put measures in place to safeguard against future, more challenging and potentially stealthy AI-enabled election threats.      

Away from AI, Meta gave details around observations of attempted foreign interference in elections. According to the Big Tech firm, Russia is still the number one source of covert influence operations seen across its platforms, with Meta disrupting thirty-nine networks since 2017.

The next most frequent sources of foreign interference were Iran (31 CIB networks), and China (11).

The company reported that the majority of CIB networks it disrupted had tried to spread their influence across many online apps, including not just Meta’s, but YouTube, TikTok, X, Telegram, Reddit, Medium, Pinterest and others, with many networks being seen to shift their activities to platforms with fewer safeguards.

Recommended reading

• What Are Tech Leaders Saying About the Election Results?
• Researchers Urge for Safeguards Against Future AI Election Threats
• How Do UK SME Bosses Feel About a Change of Government?

For example, the firm saw fictitious videos about the US elections (assessed by the US intelligence community to be linked to Russian-based influence operations) being posted on X and Telegram rather than Meta’s own apps, and when reposted to Meta platforms the content was labelled and linked to Russian actors.

The company stressed that it has taken pains to ensure the safety of Meta platforms, for example by signing the AI Elections Accord, setting up a world-first Deepfakes Analysis Unit, as well as funding and supporting the European Fact-Checking Standards Network (EFCSN) in a project aimed at improving the skills and capabilities of the European fact-checkers.

Meta also said that it continues to promote and educate citizens on their rights to vote, claiming that in the UK, its Election Day Reminder reached 43.6 million people on Facebook and Instagram, with almost 900,000 clicking through to register to vote and check their registration status.