Spanish regional elections are still nearly four months away, but Irene Larraz and her team at Newtral are already ready for impact. Every morning, half of Larraz’s team at the Madrid-based media company sets out a schedule of political speeches and debates, preparing to check politicians’ statements. The other half, who debunk misinformation, scan the web for viral lies and work to infiltrate groups that spread lies. Once the May elections are over, a national election is to be called before the end of the year, which will likely lead to a rush of online lies. “It’s going to be quite difficult,” Larraz said. “We are already preparing.”
The proliferation of disinformation and propaganda online has meant an uphill battle for fact-checkers around the world, who must sift through and verify large amounts of information during complex or rapidly changing situations, such as the Russian invasion of Ukraine, the Covid-19 pandemic, or election campaigns. This task has become even more difficult with the advent of chatbots using large language models, such as OpenAI’s ChatGPT, which can produce natural-sounding text at the click of a button, essentially automating the production of misinformation.
Faced with this asymmetry, fact-checking organizations need to create their own AI-powered tools to automate and speed up their work. It’s far from a complete solution, but fact-checkers hope these new tools will at least prevent the gap between them and their opponents from widening too quickly, at a time when social media companies are shrinking their own moderation operations.
“The race between fact checkers and those they check is an uneven one,” says Tim Gordon, co-founder of Best Practice AI, an artificial intelligence strategy and governance consultancy, and trustee of a UK charity of fact checking.
“Fact-checkers are often small organizations compared to those producing disinformation,” Gordon says. “And the scale of what generative AI can produce, and the rate at which it can do it, means this race is only going to get tougher.”
Newtral began developing its multilingual AI language model, ClaimHunter, in 2020, funded by profits from its TV wing, which produces a fact-checking show of politicians, and documentaries for HBO and Netflix.
Using Microsoft’s BERT language model, ClaimHunter’s developers used 10,000 instructions to train the system to recognize sentences that appear to include statements of fact, such as data, numbers, or comparisons. “We were teaching the machine to act as a fact-checker,” says Newtral’s chief technology officer, Rubén Míguez.
Just identifying which statements from political figures and social media accounts need to be verified is a daunting task. ClaimHunter automatically detects political claims made on Twitter, while another app transcribes politicians’ video and audio coverage into text. Identify and highlight statements that contain a claim relevant to public life that can be proven or disproved, such as in statements that are unambiguous, questions or opinions, and report them to Newtral’s fact checkers for exam.
The system isn’t perfect and sometimes flags opinions as fact, but its errors help users continually retrain the algorithm. It reduced the time it takes to identify claims worthy of verification by 70 to 80 percent, Míguez says.