Next we skipped over to “Interdimensional,” a virtual game show where we had to team up to solve a virtual puzzle. My guide hopped into a room with a large block in the middle, while I stood outside, looking through a window with an array of buttons in front of me. His job was to assess the room, and tell me which button to hit to move the block into its final goal spot. Our back and forth reminded me of completing escape rooms with friends in real life, and I could easily imagine more complex Horizon environments pushing that idea even further.

As we wound down my visit, we spend some time in Horizon’s creation mode, an empty space where you can build entire environments from scratch with your friends. In many ways, it reminded me of the white void from The Matrix — a room filled with limitless potential and creativity. Within a few minutes, my guide turned the empty space into a virtual beach, complete with sand, water and a golden sunset horizon. All the while, I spent a few minutes trying to create a palm tree. Creation mode unlocks even more menu options on your Oculus joysticks, and you can also make yourself giant or tiny to build out individual components. I’m not much of an artist, but the variety of tools available, and the deep amount of customizability per item, could lead to some truly interesting environments.

Horizon sounds like a social VR dream. But the reality may be much more difficult than Facebook realizes, especially when it comes to security. We’ve seen this story plenty of times: Give people a public space on the internet, and harassment and abuse inevitably follows. The company says moderators will be alerted if people end up reporting, blocking or muting a specific user, or if they notice any other strange behavior. Those mods will be able to listen in on your VR experiences (remember, everything is public), and they also have access to a buffer of recorded world data (though it’s unclear how much data that entails). They’ll be able to ban users outright if they find any violations.

But, really, that’s all just common sense. The real question is how Facebook could help prevent harassment from happening in the first place. Just like going anywhere in public, entering Horizon is a risk. We don’t know how many moderators Facebook will actually have once the service opens up. And it’s hard to tell if these safety tools will be enough if there’s a huge disruptive event, involving lots of bad actors. Clearly, Facebook is still trying to figure all of this out. And until it does, I don’t expect Horizon to leave beta anytime soon.