Facebook Inc. has been plagued by fake news appearing in people’s news feeds for quite some time, but on Tuesday the company itself was fooled by false reports of a bomb attack in Bangkok, Thailand.
The fake news report tricked Facebook’s algorithms into activating its Safety Check feature, and thousands of users in Bangkok marked themselves as “safe,” despite the fact that there was no explosion.
Facebook’s Safety Check feature has proved to be very useful during natural disasters, terrorist attacks and other emergencies, allowing users to quickly let their loved ones know they haven’t come to any harm. To date, the feature has been activated more than 335 times. But on Tuesday the tool malfunctioned and pushed out a false alert about an explosion in Bangkok, before being deactivated after about an hour.
It was no surprise that thousands, including this author, were fooled by the alert. The warning cited “media sources” as confirmation, and even linked to a website called Bangkok Informer that had posted a video of the 2015 Erawan Shrine bombing to back up its false claims.
In a statement to The Verge, Facebook said that the false link was automatically populated by Safety Check, even though no recognized news organizations were reporting the news. It also claimed to have “verified” news of the blast with an apparently trusted source – though of course that source proved to be incorrect.
“As with all Safety Check activations, Facebook relies on a trusted third party to first confirm the incident and then on the community to use the tool and share with friends and family,” a spokesperson said.
Reports suggest that a protest outside Bangkok’s Government House may have been the real source of the bomb rumors. According to the BBC’s Thai language service, one protester allegedly threw “ping pong bombs,” which were described as “giant firecrackers.” No injuries or explosion were reported, but the commotion did cause some alarm for passersby.
The false alarm comes after Facebook said in November that it would begin automating Safety Checks rather than activating them on its own. The company now relies on an algorithm that scans user activity to create alerts, tracking keywords that indicate danger such as “shooting,” “explosion,” “bombing” or “earthquake” that pop up in highly populated areas.
Clearly, that algorithm still has room for improvement.