I imagine that the building was remote and likely marked as a suspected terrorist site. The person who posted the photo was likely flagged as a terrorist based on previous data and was identified by facial recognition in the photo. Having a known terrorist at a suspected terrorist facility was likely enough justification to label the site as malicious and bomb it.

The implications of something like this are interesting - it's just a graph theory problem, and hypothetically the computer could be programmed to automate a drone strike when a location reaches a certain score threshold.