Facebook Alters Safety Check to Improve Accuracy

After several recent missteps with its Safety Check feature, Facebook today announced that it would make changes to how members are notified that friends in disaster areas are safe.

During a disaster like a flood or mass shooting, Facebook will take note of which users in the affected area use Safety Check to mark themselves safe, and then prompt their friends who may be nearby to also check in using the feature. That could help avoid situations like a bombing in Lahore, Pakistan, in March, during which Facebook accidentally prompted users who were nowhere near the city to check in safe.

The changes are part of an ongoing effort to make Safety Check more responsive to the needs of a community in crisis. Facebook was criticized after it activated the feature for bombings in Paris last fall, but not for other similar bombings in the Middle East. In response, CEO Mark Zuckerberg announced that the feature would be rolled out to all major disasters.

That has meant a much more frequently activated service. Between January and May 2016, Safety Check was turned on 17 times, compared with 11 times in 2014 and 2015 combined, according to Facebook engineer Peter Cottle.

"As our efforts ramped up, the team began solving the technical challenges associated with more consistent and frequent activations," Cottle explained in a blog post. "To create a system that we can launch anywhere at a moment's notice, we scaled our infrastructure to handle larger events more efficiently and automated many of the manual steps previously required for activation."

The user-initiated check-in prompts announced today are a significant modification to that automated process, which may have contributed to the mistaken notification prompts sent to users in areas unaffected by crisis.

Cottle wrote that the company will be testing more "new ways for the community to initiate and spread Safety Check" in the next few weeks.

This article originally appeared on PCMag.com.