Facebook introduces new photo-matching tech to fight revenge porn

Revenge porn is one of the most insidious problems on social media and the internet at large — and Facebook is now taking new steps to fight it on one of the world's largest online platforms.

The social network announced it's rolling out a new set of tools for users on Facebook, Messenger, and Instagram to prevent their most intimate images from being shared without their permission. The system won't immediately block problematic images from being posted — but once an image has been flagged and its subject identified, photo-matching tech will be used to make sure it won't be uploaded again.

The tools and new protocols were developed by Facebook through a partnership with safety experts and advocacy organizations, including the National Network to End Domestic Violence, Center for Social Research, the Revenge Porn Helpline (UK) and the Cyber Civil Rights Initiative.

There are multiple steps to address revenge porn under the new rules. Users are encouraged to report images of themselves and others that appear to have been shared without the subject's permission. Facebook made a guide to lead victims through that process, which you can find here.

Once a photo has been reported, specially trained reps from Facebook's Community Operations team review the pic and remove it if it violates the platform's community standards. This is where things might get tricky: an intimate photo posted on Facebook's public network doesn't need to be "explicit" under Facebook's definition of the term to humiliate or traumatize its subject.

That's why the new revenge porn policies won't depend on Facebook's increasingly advanced AI-based photo recognition tools and will turn instead to a human moderator's judgment. Antigone Davis, Facebook's Head of Global Safety, told TechCrunch that human discretion is still needed for such a sensitive issue. "There is significant context that’s required for reviewing non-consensual sharing," she said.

Profiles of the users that post removed images will be disabled "in most cases," although an appeals process will be available.

After images have been flagged and removed from any of the platforms, photo-matching tech will prevent them from being reuploaded. If anyone tries to share the pic again, they'll be blocked from sharing it and alerted that it's in violation of community policy.

That raises another potential issue: Could there be a way to warn the subject of the image that someone is trying to post without their permission a second time? Facebook reps said that type of functionality isn't in the cards for now, and pointed us to an all-in-one guide Facebook developed with the Cyber Civil Rights Initiative and other companies to help victims with next steps.

The social network is also touting new resources for victims of revenge porn posts through partnerships with the aforementioned safety organizations.

The United States Marine Corps was recently shook by a photo sharing scandal after a secret Facebook group that shared a Google Drive link filled with explicit photos of servicewomen came to light. Since many of those photos weren't directly posted to Facebook, the new tools might not have prevented the group's initial impact — but the increased resources for support tools give those victims, and others in similarly troubling circumstances, even more places to turn in its aftermath.

Revenge porn isn't the only area Facebook is stepping up to prevent harmful behaviors on the platform, to its credit. Last month, it expanded and strengthened its suicide prevention tools, which could potentially depend on AI systems to ID users who might be at-risk of self harm.