“We are all at risk of manipulation online right now.”
So begins a short animated video about a practice known as taking content out of context and how it can be used to misinform people online. The video identifies signs to watch out for, including surprising or unusual content, seemingly unreliable sources, or video or audio that appear to have been manipulated or reused in a different way.
Though it may not look like it, this 50-second video is actually an election ad—one of three that Google will be sharing across five European countries next month before the European Union’s June parliamentary elections. But unlike traditional election ads that are designed to persuade people how to vote, these are seeking to educate voters about how they could be misled. It’s an initiative that Google describes as preventatively correcting misunderstandings—or, more simply, “prebunking.”
“It works like a vaccine,” Beth Goldberg, the head of research at Google’s internal Jigsaw unit, which was founded in 2010 with a goal to address threats to open societies, tells TIME. By enabling prospective voters to recognize common manipulation techniques that could be used to mislead them—such as blaming others or creating division—Goldberg says that prebunking “helps people to gain mental defenses proactively.”
Concerns about AI-generated misinformation and the impact it stands to have on elections around the world this year’s election season. This is particularly true in the E.U., which recently passed laws requiring tech firms to increase their efforts to reduce misinformation amid concerns that it could distort the results.
Contrary to what one might expect, prebunking ads aren’t overtly political nor do they make any references to any specific candidates or parties. In the video about taking content out of context, for example, viewers are shown a hypothetical situation in which an AI-generated video of a lion let loose on a town square is used to stir up fear and panic. In another video, this time about blaming others, they are shown an incident in which a community lays sole blame on another group (in this case, tourists) for the litter in their parks without exploring other possible causes.
The beauty of this approach, Goldberg notes, is that it needn’t be specific. “It doesn’t have to be actual misinformation; you can just show someone how the manipulation works,” she says, noting that keeping the content general and focusing on manipulation strategies, rather than the misinformation itself, allows these campaigns to reach people regardless of their political views.
While Google’s prebunking campaign is relatively new, the tactic is not. Indeed, the concept dates back to the 1960s, when the social psychologist William McGuire sought to understand people’s susceptibility to propaganda during the Cold War and whether they could be defended against it. This culminated in what McGuire called “inoculation theory,” which rested on the premise that false narratives, like viruses, can spread and that by exposing people to the facts first, they can become less susceptible. But it wasn’t until decades later that the theory began being applied to online information. In recent years, Jigsaw has conducted prebunking initiatives in Africa and Latin America. Its forthcoming European campaign, which formally kicks off in May, will primarily be shared as short ads on YouTube and Meta platforms targeting voters in Belgium, France, Germany, Italy, and Poland. Afterwards, viewers will be invited to take a short, multiple-choice survey testing their ability to identify the manipulation technique featured in the ad.
Whereas prebunking doesn’t necessarily face as much resistance as more conventional forms of combating misinformation such as fact checking or content moderation, which some critics have compared to censorship, it isn’t a perfect solution either. Jon Roozenbeek, an assistant professor in psychology and security at King’s College London who has spent years working with Jigsaw on prebunking, tells TIME that one of the biggest challenges in these campaigns is ensuring that the videos are captivating enough to hold viewers’ attention. Even if they do, he adds, “You can’t really expect miracles in a sense that, all of a sudden after one of these videos, people begin to behave completely differently online.” he says. “It’s just way too much to expect from a psychological intervention that is as light touch as this.”
This isn’t to say that prebunking doesn’t have an impact. In previous campaigns, post-ad surveys showed that the share of individuals who could correctly identify a manipulation technique increased by as much as 5% after viewing a prebunking video. “We’re not doubtful that the effect is real; it’s just you can argue over whether it’s large enough,” Roozenbeek says. “That’s the main discussion that we’re having.”
While Jigsaw has led the way on prebunking efforts, they’re not the only ones utilizing this approach. In the U.S., the Biden administration has sought to counter Russian misinformation in part by declassifying intelligence forecasting the kinds of narratives that it anticipated the Kremlin would use, particularly in the run up to Moscow’s 2022 full-scale invasion of Ukraine. This practice has since extended to China (where the U.S. government used declassified materials to forecast potential Chinese provocations in the Taiwan Strait) and Iran (the U.S. declassified intelligence claiming that Tehran had transferred drones and cruise missiles to Houthi militants in Yemen that were being used to attack ships in the Red Sea). What the White House has billed as “preemptive disclosure” is just prebunking by another name.
Working with academics and civil society organizations across the E.U.’s 27 member states, Jigsaw’s latest prebunking campaign is set to be its biggest and most collaborative effort yet. And in an election that will see hundreds of millions of voters go to the polls to elect what polls project could be the largest turnout ever, the stakes couldn’t be higher.