The Mountain View, Calif.-based company has used these fact-checking labels for years in its main search results and on video-streaming site YouTube. In December, Google said fact-checks appear more than 11 million times each day in search results.
“Photos and videos are an incredible way to help people understand what’s going on in the world,” Google product manager Harris Cohen wrote in a blog post announcing the new fact-check labels. “But the power of visual media has its pitfalls — especially when there are questions surrounding the origin, authenticity or context of an image.”
Tech companies’ efforts to fact-check the myriad claims made across their sites have become a major focal point as advertising and campaigning for the 2020 U.S. elections heats up. Twitter’s decision to label two of President Trump’s misleading tweets about mail-in ballots with fact-check links set a de facto standard for social media companies last month. Twitter labeled another tweet by Trump with a warning last week after the president tweeted a doctored video showing fake CNN headlines. The company said it violated its policies on manipulated media.
Fact-checking from social media and other tech companies has become common in the past three years — Facebook, Twitter and Google all do it to some extent — but it is by no means universal and often relies on news media and other partners to publish a fact-check and make sure the companies see it. It can also be applied unevenly, something that triggers complaints.
YouTube CEO Susan Wojcicki noted in a Washington Post Live interview last week that the company removes videos that violate its policies — which prohibit some hate speech, inciting violence and some incidents of manipulated media that could cause disinformation — whether they are from a politician or anyone else. But the company keeps some of the videos on the site if they are presented in context, by a news report or for educational purposes, she said.
Google’s efforts are the first widespread initiative to try to fact-check images, said Nina Jankowicz, a disinformation specialist at the Wilson Center and author of the forthcoming book “How to Lose the Information War.”
Seeing manipulated images and video can be a lot more convincing to people than disinformation in text, Jankowicz said, and she’s hopeful that Google’s labels will at least cause people to think before they post. It won’t work for everyone, she said.
“Some people will just push back against any content they believe is not true,” she said. But others may pause. “It could get people to slow down and think before they share.”
Manipulated photos are one tool to spread misinformation. This month, Fox News published manipulated photos of a protest zone in Seattle, making it seem as though a city block was on fire. But it was really multiple images edited together, and the fire was actually in Minnesota.
Google image search does not yet show a fact-check for the Fox News photos, which Fox has since removed. The company does not apply fact-checking labels to all manipulated photos. It instead points users to articles that have fact-checked the photos, if they exist. It is using the service ClaimReview, which organizes fact-checks throughout the Internet and makes them visible to search engines.
The company used the example of an image showing a giant shark swimming along a Houston street. Now a search for the shark image — which was edited to make it seem as though a storm had caused the ocean wildlife to swim alongside cars — will show a small fact-check label next to a photo attached to a PolitiFact article.
Google said it is launching the feature fully this week.