Humans are more to blame than bots for the spread of fake news on Twitter.
A study by the Massachusetts Institute of Technology published in the academic journal Science found that fake news was more 70 per cent more likely to be shared on the social media platform than accurate stories.
Sinan Aral, who supervised the study, said: "False news spreads further, faster and deeper than the truth in every category of news."
The study is the largest ever analysis of the spread of false information on Twitter, covering three million accounts and 4.5 million tweets between 2006 and 2017.
Around 15 per cent of the accounts monitored were bots, but when they were removed from the data, fake news was still more likely to be shared than the truth.
Aral believes the spread of false news by Twitter uses can be attributed to users' desire to stand out from the crowd.
The study also found that tweets about truthful stories only rarely reached an audience of more than 1,000 people, while the top one per cent of false news stories routinely reached between 1,000 and 100,000 people. And accurate tweets spread much slower than fake news, taking on average six times as long to reach an audience of 1,500 people.
Of all the different types of news analysed, the study found that political news was more likely to go viral than any other kind. Conspiracy theories that Barack Obama wasn't born in the United States and that Hillary Clinton was seriously ill during the 2016 presidential election were particularly prominent.
While this is the largest study of fake news on Twitter, it's much harder to infiltrate Facebook data to monitor the spread of false information on that platform.