Is the world ready for virtual graffiti?

Imagine a world that’s filled with invisible graffiti. Open an app, point your phone at a wall, and blank brick or cement becomes a canvas. Create art with digital spraypaint and stencils, and an augmented reality system will permanently store its location and placement, creating the illusion of real street art. If friends or social media followers have the app, they can find your painting on a map and come see it. You might scrawl an in-joke across the door of a friend’s apartment, or paint a gorgeous mural on the side of a local store.

Now imagine a darker world. Members of hate groups gleefully swap pictures of racist tags on civil rights monuments. Students bully each other by spreading vicious rumors on the walls of a target’s house. Small businesses get mobbed beyond capacity when a big influencer posts a sticker on their window. The developers of Mark AR, an app that’s described as “the world’s first augmented reality social platform,” are trying to create the good version of this system. They’re still figuring out how to avoid the bad one.

Mark AR is one of the first projects built on Google’s Persistent Cloud Anchors. The app, created by mobile publisher iDreamSky and Subway Surfers developer Sybo, debuted at last week’s New York Comic-Con, where visitors could borrow a phone and walk through a Mark AR pop-up installation, either viewing professional artwork or creating their own. Talented artists might use a virtual spraycan to paint freehand. Everyone else (including me) could pick from a set of comics-themed stencils. In the future, users could make their own stencils or even design images in Photoshop and import them directly.

Mark AR’s creators are planning more pop-up exhibits, and after these small trial runs, they plan to test the app in a single city. “A city launch will be where we’re testing: can we handle the moderation? Can we make sure people are playing safe?” says iDreamSky president Jeff Lyndon. “Once we can handle a city, we know exactly how we can scale our business to a national launch.” But if existing social networks have taught us anything, it’s that these are massive, complicated — sometimes even impossible — questions to answer.

At launch, Mark AR is supposed to work a bit like Facebook. Users will log in with real names, probably through Facebook itself. When they create art, they can share the location with a single person, a list of friends and followers, or the members of a group. Google’s ARCore platform stores the location using GPS and computer vision, capturing details in the environment to use them as anchor points. When somebody shares art with you, a thumbnail will appear on a map; if you visit that location and point your phone at the place shown in the thumbnail, you’ll see whatever image they’ve created.

There are lots of potential technical issues, since Cloud Anchors are still very new. But social topics — like sharing, privacy, and abuse — are more interesting. Pokémon Go, the first successful AR game, raised its share of unexpected questions. Should people be able to catch pokémon in the Holocaust Museum? (No.) Could it be illegal to place a digital marker on a person’s real house without permission? (Unclear.) Should app makers worry about their users falling into ponds? (Apparently.)

Mark AR faces these issues plus the complications of running a creative platform where anybody can upload content. They’re also one of the first players trying to launch this kind of network as a mainstream product — although Microsoft will face similar issues with its AR game Minecraft Earth, something the company readily acknowledges.

Mark AR’s creators are taking some cues from Pokémon Go — they’re going to geofence physical spaces like memorials to be off-limits, for instance. And they hope a real-name policy and the friend-based model will limit people making offensive or harassing images. “Because there’s no anonymity, that helps govern what people are doing,” says Sybo CEO Mathias Gredal Norvig. (It’s unclear how true that is — Facebook has faced repeated problems with closed groups devoted to swapping non-consensual pornography or degrading women or immigrants.)

Lyndon adds that Mark AR will devote resources to addressing abuse. “We are working on hiring a human moderation team, and we are also working with a few tech companies to provide image recognition — to provide kind of a moderated machine learning, to just go through some of the images very quickly.” That’s similar to the approach that bigger, purely online social networks have taken. It’s been difficult to scale, though. AI can’t make sophisticated moderation decisions, and human teams are often overworked and sometimes traumatized by constantly viewing awful content. And while iDreamSky and Sybo are both well-established companies, they don’t have the kind of resources that, say, Facebook could throw at the problem.

Just to be clear: Mark AR isn’t equivalent to somebody tagging a building with real graffiti. Users can’t deface or cover up each other’s work. People will have to seek out the digital art. The company can remove paintings at any time. And Mark AR might not end up being much of a public platform. Its creators tout options like letting people decorate private rooms or create scavenger hunts for friends and family. If that happens, moderation might not be a huge issue, assuming the app catches on at all — which is far from a sure thing.

But Mark AR’s whole purpose is imitating a public art form. So its most interesting uses involve, well, people making public art. It’s a natural fit for events like comic-cons and music festivals, where visitors come together in a physical space for creative purposes. And the idea of wandering around a city, finding the random tags people have left behind, is fascinating. Can Mark AR build a new augmented reality without falling into the same traps as our current digital worlds? That’s a fascinating question, too.