Is it possible to stop the harmful spread of false news on the Internet? British researchers believe so. They have developed an AI that can stop Fake News before it spreads on the web.
“Fake news” is an unattractive term that gained notoriety primarily due to former U.S. President Donald Trump and usually means false reports. In this case, “news” is manipulated in such a way that readers are given the impression that it is genuine.
Often, however, they are erroneous, deliberately misleading or simply made up. The problem is that because they appear so credible, they spread rapidly on the Internet – especially via social media. Tragically, such false reports spread even faster than real news.
Once in circulation, fake news can hardly be stopped
Afterwards, it is virtually impossible to spread the corrections of these fake news just as well or even to show users that the information is not true at all.
To stop the spread of fake news, it is therefore necessary to act very quickly.
That’s why researchers at University College London have developed an effective “early warning system“. Behind it is a self-learning algorithm that can detect these false reports in real time and sound the alarm before they spread.
Early warning system for false reports
The smart algorithm focuses on domains that are most likely to be the source of such false reports.
In particular, the registration data of these domains is very revealing, says one of the study authors Anil Doshi. For example, anonymously registered domains are more likely to be platforms from which fake news emanates, he said.
“By using domain registration data, we provide an early warning system because we are using data that is difficult for stakeholders to manipulate. The actors who produce false information tend to want to stay hidden. We take advantage of that in our model.”
Apparently, the artificial intelligence (AI) programmed in this way can detect fake news channels very reliably. The tool was able to correctly match 92 percent of fake news domains and also 96.2 percent of domains with non-fake information.
Forecast: It’s getting worse
The starting point for the tests were domains registered in connection with the 2016 U.S. elections.
The researchers believe their AI can help authorities and platforms detect fake news. These could then not only stop fake news, but also block the domains.
Realistically, the researchers believe that the spread of such fake news will always increase before elections. In their estimation, this will become more rather than less in the future.
Stop Fake News: Platforms must also act
That’s why they are also making clear demands of social media platforms and digital companies: They should invest in smart technologies themselves to stop the spread of fake news.
The tool developed by the British researchers currently works very well. But it is only a matter of time before the relevant players adapt their behavior. Research must therefore continue to stay on the ball.