Facebook announced that it will require users to agree to activate facial recognition features on the site. That includes usage of Facebook’s ‘tag suggestion’ feature, which automatically suggests a user’s name as a tag on an uploaded photo.
According to Facebook, users who log in the social network today will be given information about tag suggestion and facial recognition, as well as the option to turn the feature on. If users take no action, the feature will be deactivated by default. Prior to this, the feature had been activated by default, displaying the user’s name and notifying the user when Facebook’s A.I. detected their image in a photo or video.
It will still be possible for Facebook users to manually tag people in photos, even if their facial recognition features like tag suggestion have been turned off.
This new “opt-in” policy requires Facebook users to give consent to features that could compromise their privacy. Opt-in policies have often been held up as a best practice by privacy advocates, since they give users the most privacy by default, rather than requiring that they take action to turn features off. Europe’s GDPR privacy rules followed this logic, for instance, when requiring opt-in consent for websites’ use of tracking cookies.
Facebook’s suggest tag feature, powered by automated facial recognition, was first activated for all of the social network’s users by default in 2011. It was halted in the EU in 2012 over privacy concerns, but Facebook reactivated it in 2013. Starting in 2017, a user was actively notified when a photo of them was uploaded, which some worried could open the door for harassment.
Today’s announcement is a step to fulfill Facebook’s ‘pivot to privacy,’ announced with much fanfare in March, part of a reaction to a drumbeat of disastrous revelations about the company’s mishandling of user data. It was also motivated, according to CEO Mark Zuckerberg, by a strategic belief that users will increasingly gravitate to services that offer more privacy.
The move follows a parallel decision by Apple last week to Siri voice recordings temporary by default, and ask users to opt-in to having them saved. Apple has scrupulously positioned itself as a more privacy-conscious alternative to the likes of Facebook and Google.
The iPhone-maker can afford to put privacy first, because it’s a hardware maker, and not reliant on customer data to turn profits. But opt-in policies have real costs to digital media companies that depend on advertising. It’s likely that a large number of Facebook users will never opt in under the new policy, with several likely effects.
Facebook uses user-contributed data to train its image recognition algorithms, so collecting fewer tags from users may slow improvements to its tools. In addition, Facebook’s new policy will reduce the number of notifications that opted-out users receive, possibly reducing time spent on the social network and, in turn, lessen amount of advertising they view.
The new policy could also reduce the data Facebook can gather about users who don’t opt in. That includes not just data about their appearance, but other things, such as their connections to other users. One study found that a European privacy rule reduced the effectiveness of online advertising targeting by 65%. That large of an impact would seem likely to reduce advertising sales, Facebook’s primary source of revenue.
More must-read stories from Fortune:
—How Reliance Jio became India’s wireless wonder
—Deloitte’s plan for fighting employee burnout: let A.I. take over the dreaded HR and IT tasks
—We’re in a data literacy crisis. Could librarians be the superheroes we need?
—No humans needed: Chinese company uses A.I. to read books and the news
—ProPublica: How Amazon and Silicon Valley seduced the Pentagon
Catch up with Data Sheet, Fortune’s daily digest on the business of tech.