If a picture is worth a thousand words, then the Internet is the world’s longest saga, with billions of images plastered across billions of pages. With this advent came both unprecedented access to information and never before seen accessibility to tell one’s story to the world, with social media platforms turning into legitimate sources of information for journalists.
But a caveat has arisen in the form of fake images.
Honest mistakes and malicious subversion play into the hands of those who wish to incite hate. In May, the New York Times mistakenly used a picture of a Pakistani girl to represent Rahaf Al-Masri, a young girl killed in the Israeli-Palestinian conflict, on a collage dedicated to Palestinian children who had lost their lives in that long-drawn saga.
Detractors flocked immediately to social media, with some using the mistake as proof that the death of any Palestinian children was all a fabrication.
One comment on Facebook read: “The Left and Hamas do this ALL the time.” Some commentatators felt that no Palestinian children were being killed at all, so that single mistake was definite proof of a larger conspiracy theory to absolve Hamas (a Palestinian nationalist militant organization) of any wrongdoing.
A mistake that was painfully avoidable.
The remedy for the future is to only use images that are verifiable. But such advice is easier said than followed: there are millions of fake images on the Internet. And an image need not be obviously photoshopped to be fake. Real photos can be used to back up false claims, making it all the more tricky to sift the truth from the flood of misinformation.
To Sam Dubberley, speaking at Amnesty International’s Journalism Under Attack Conference on May 21, media verification is akin to a jigsaw puzzle – little clues that, when carefully pieced together, can help one judge if something is real or fake.
But not all videos he receives are real. Sometimes, images of an older atrocity are reused as proof of a newer one. Dubberley recounts a case where an asylum seeker wanted to use a video alleged to be of a Libyan human trafficking complex. But the video was not from Libya, but Nigeria – a man in the video was seen wearing traditional garb from the West African nation.
Fake footage can hurt real causes, so journalists and human rights activists alike must exercise caution. Fortunately, there are methods and tools to help us navigate the increasingly murky waters of the visual side of the Internet.
Dubberley lists some questions that journalists can ask to determine whether a piece of media is real or fake, such as the following:
Do you know who the videographer or photographer is?
Who uploaded the video?
When was the media piece captured – date/time?
Where was the photo or video taken?
Can the media piece be corroborated?
What is the motivation behind the video?
Some media pieces are easier to track than others. For instance, to determine whether a piece of media is original, a quick reverse image search usually does the trick.
There are several search systems that can be used to run a picture or video (not all support video verification). Some that Dubberley mentioned include Yandex, Tineye, Google, Bing, and Invid. The databases in these search systems are different, so you should run a piece of media through all of them just to be sure. If the searches bear results, that means that the image or video has been used before and is not original.
The more complex part of verification comes when an image or video appears to be original. Dubberley says this situation can arise when a piece of media is said to be proof of a claimed event.
In such cases, the verification process is more complex than just simple search engine plugins.
Dubberley recounts how he verified the time of capture for a video of a massacre using a few frames of one of the perpetrators. As the man was wearing a jersey of Liverpool football club, he checked that the player whose name was plastered across the back – Mohamed Salah – only moved to Liverpool in 2017 from A.S. Roma football club. That indicated that the said video was created after 2017, strengthening the claim that it showed a recent massacre.
Dubberley said that details count towards determining the exact location and time of an event. The changing of graffiti or shadow lengths can place a video or photograph down to the exact day and hour. His team uses Photoshop to piece together large panoramas, which are then corroborated using Google Earth Pro to locate the exact area where a piece of footage is shot.
But freelance photojournalist Tom White said that the average journalist and/or reader has neither the luxury of time nor the will to verify every single piece of footage he or she receives.
White acknowledges that safeguards to stem the tide of manipulated images are available. Fact-checking sites, like Politifact are dedicated to debunking false news – images included – online, but their effect is limited.
“Relying on a regulatory body, or even technology (in the form of fact-checking AI) is misplaced faith,” he says.
White calls on the average person to be more responsible, adding that average citizens need to ask more questions about the news that they receive. A media-literate and educated public is key to debunking fake images – and by extension, fake news.
“Education is key. Editors need to be educated, publishers need to be educated. We all need to be part of that chain of trust and chain of verification,” adds White. In an increasingly polarised world that is seeing trust in journalism fall, it is more critical that news organizations only use verifiable images to help build the chain of trust. One way to do that is to ensure that the footage used for each story is real – context and all.