Facebook is giving the power of Safety Check to the people.
The social media giant said Thursday it is no longer manually activating its crisis response tool. Instead, it will rely on its 1.2 billion daily visitors to help automatically launch Safety Check.
Safety Check sends a message to Facebook users in areas of immediate danger, allowing people to notify friends and family that they are alive and well. It has been used for natural disasters like hurricanes, as well as mass shootings and terrorist attacks.
But Facebook has also been criticized for failing to activate Safety Check in some parts of the world. Safety Check was activated for the Islamic State group attacks in Paris last November, but not the day before for suicide bombings in Beirut. Facebook CEO Mark Zuckerberg responded at the time by saying the company would activate Safety Check more widely. "We care about all people equally, and we will work hard to help people suffering in as many of these situations as we can," he wrote in a November 2015 post.
In the past two years, Facebook turned on Safety Check 39 times. Compare that to 335 dangerous events flagged by its community-based Safety Check tool since the company began testing it in June. One of the first instances of a community-generated Safety Check was the Orlando nightclub shooting in June.
The community-based Safety Check tool, which is now live, works by picking up keywords indicating danger concentrated from certain areas, like "earthquake," "fire" or "shooting." Facebook then verifies there is a real threat with a third-party security company before letting the Safety Check roll out. From there, users can choose to share the Safety Check with their friends and family in the same area.
The shift allows the community to decide the urgency of the nearby danger, something Facebook has struggled to grasp.
When Facebook had control of Safety Check, it had a high standard of what counted as a disaster, Peter Cottle, the social network's lead engineer on crisis response, said Thursday. He was speaking at the company's Social Good Forum in New York. Facebook only launched a Safety Check for about 11 percent of community activations.
By localizing Safety Check, the community can decide whether an event should be considered an emergency.
"A typhoon in the Philippines might have six inches of water in your house, and in California, that'd be a big deal," Cottle said. "But in the Philippines, we did research there, and people said this wasn't a big deal."
If users ignore the Safety Check prompt, Facebook sees it as the event not being as urgent of an emergency and the tool will eventually fade away.
"We can tell how many people are spreading this and marking themselves safe, and how quickly it's growing," Cottle said. "There's a real strong measure of urgency based on the rapidness of the people who are using the tool."
Facebook is also testing out a Community Help page that users can access after checking in as safe. There, users can post if they need shelter, food or supplies, or if they can provide any of those resources.
Cottle said the social network was inspired to create this feature after seeing the outpouring of help from people on social media, like the trending #PorteOuverte ("Open Door") hashtag offering shelter following the Paris attacks.
Facebook wants to put all the help in one centralized area, with features like GPS location and checking in to let people know you've been helped.
"What we saw is the inefficiencies in matching were so strong. You couldn't find people nearby," Cottle said. "Some of these tweets may go viral, and they've already been helped, but people are still retweeting it."
The Community Help feature is expected to be available to the public by January.