Google is introducing new features in Google Photos to help you share content with friends and family.
One new feature, called Suggested Sharing, uses software to make recommendations about which photos to share and with whom, based on who is in the photos and who you tend to communicate with in the Google Photos app and in Gmail.
Shared Libraries, another new feature, will allow you to set up automatic sharing of photos featuring certain people or starting at a certain date with one chosen contact.
“You don’t have to use any of these features,” Google Photos product lead David Lieb, who is responsible for design and product management of the app, told Recode. “And if you do choose to use them, we’ve provided a bunch of controls.”
The features may be new, but the question they raise is a familiar one for Google: Where do you draw the line between useful and creepy?
Google presents the new features in the light of them being useful. Of course, your spouse would love to automatically receive every photo you take of the kids. Of course, it would be nice to have some help selecting photos and distributing them to friends after a birthday party or bowling meetup.
And in these examples Google’s computer vision technology works flawlessly, identifying the right faces and suggesting you send photos to the right people.
But our photo libraries aren’t as idyllic as Google presents them to be. Technology doesn’t work perfectly, and people make mistakes. Here’s what could happen:
- You set up photo sharing with your spouse or partner. The two of you split and you both forget to turn it off. They continue to receive whatever photos, and you may wish you hadn’t sent those or they may not wish to receive them. You can go still switch off the connection, but you can’t un-see the photos.
- A demanding, and perhaps abusive, friend or family member insists that you turn on photo sharing. It becomes more difficult in an already difficult situation to maintain your privacy. Lieb notes that whoever you share photos with does not see an indication of whether you have the app set to show your selected sharing contact all of your photos or just a subset.
- You’re taking photos of a group of friends and Google is suggesting you send the photos to someone you don’t want to show them to for whatever reason. Perhaps the suggested recipient is an ex or a former friend with whom you not longer speak. It’s your decision whether to press send, but this can be an unsettling experience, and sometimes fingers slip.
- Maybe that person’s name is included in a group of names, you miss that it’s there or mistake it for another name, act quickly, and hit send.
- You take a photo you don’t want to share with anyone. Google either automatically shares it or suggests that you share it.
You get the picture.
Lieb pointed out that users have options in Google Photos to turn off recognition of their face and prevent their names from appearing automatically in the Suggested Share field of other users’ Google Photos apps.
Interestingly, he said the new features won’t deepen Google’s understanding of the social connections among users. They rely on what Google already knows.
And Google knows a lot, as the features will constantly remind you.
We’ve seen these situations, or their potential, in other applications of artificial intelligence and automation by Google. Examples include location tracking in Maps and the ability to access other users’ private information through voice technology.
Maps allows users to share their location with other users for set periods of time or by default. In abusive relationships, this can open an opportunity for one person to demand constant access to the whereabouts of another.
Google Home can now recognize multiple voices, but that technology is still being perfected and can be fooled by voices that sound similar, or by recordings. This opens up the potential for users to access each other’s calendar appointments and other private information.
What we’re learning from these new opportunities for invasion of privacy is that maybe it’s not an algorithm’s understanding of us that we need to fear, but its misunderstanding. You can throw a lot of data at a computer and it still may not pick up on important social context. Even humans can be tactless.
This article originally appeared on Recode.net.