In June, Elon University launched its first AI Play camp. In one activity, teams of middle school students played Tic-Tac-Toe, not as a child might, but rather, like a machine. Tasked with writing and then following a set of fixed instructions, the students play the game as an AI agent would. That experience of following instructions and playing without improvisation is becoming surprisingly commonplace. Not by choice, but by default. The games and creative tools kids use to play with are now being played for them.
With a prompt, AI can now generate a Roblox universe, create an epic build in Minecraft, finish a half‑drawn character, or turn a rough story idea into a finished one.
As generative artificial intelligence embeds itself more deeply into childhood, from toys to creative tools, from school to sports, concerns have mostly focused on risk. We worry about young people’s exposure to harmful content and risky social interactions with companion chatbots. We worry about the atrophy of their independent reasoning and problem-solving skills as students rely on AI more and more to write and plan for them.
At the same time, we also worry that we are not doing enough to prepare our children for an AI-enabled future.
While direct evidence on youth-AI interaction is still emerging, it is safe to say that all of these concerns are real.
But the focus on risk is incomplete. It overlooks something quietly disappearing: the space where young people can freely explore possibilities that have not yet been selected, optimized, or created for them. That space has always had a name. We used to call it play.
If you want to see what’s at stake, you don’t have to look far. My tween daughter and her friends live in the overlap between the space of play as it used to be and what it is becoming. In one, AI finishes their drawings and completes their stories, then feeds them an endless stream of AI-generated short videos.
In the other, they have spent almost a year building (and breaking) a shared world in a group chat, an unruly, ever‑changing playspace filled with inside jokes, memes, and rules that make no sense to anyone outside it, least of all me. No adult designed it. No chatbot tells them what to say next. It exists because five girls keep showing up to create and play freely together. They have power there because the play belongs to them.
In 2013, the UN established the right to play as a fundamental right for all children, essential to their development and wellbeing. But AI threatens to take this right away by flattening play into a sequence of suggestions to accept or reject, rather than opening it up into a possibility space to imagine and explore.
This flattening and ordering of play, however, is not a forgone conclusion. I see young people exercising their right to play in many of the online playspaces I study: Minecraft servers, Discord communities, and other peer‑to‑peer networks. In a recent book chapter, I draw on more than 15 years of research and design experience to outline principles for creating online play communities where young people can thrive. In these play communities, young people build moderation systems, mediate conflicts, and create norms for their communities, sometimes with the help of AI, but more often through play.
For a young person who is stuck or impatient, AI can feel like relief or even magic. In one study, children described generative AI as a kind of creative accelerator, helping them turn a stuck idea into a finished picture or plot in seconds. This can feel thrilling and even reassuring when you aren’t sure about what to do next. But when AI takes over the initiating, imagining, and finishing on a young person’s behalf, it usurps vital developmental work that belongs to young people themselves.
If we want to protect the right to play online in an age of AI, we have to create more opportunities for free, self-directed play. That means saying “yes” more often to the kinds of messy, half‑visible online playspaces adults don’t quite understand. This means letting our children play Minecraft on a server with neighborhood friends or sponsoring a YouTube channel where they can make and post their own videos. It means protecting unscripted time in group chats, online games, and creative online spaces like DeviantArt or Archive of Our Own (AO3). It means resisting the everyday use of AI tools that take on the work of ideation, explanation, reassurance, or completion.
Treating play as a right means designers and policymakers need to think beyond risk. We need developmentally aligned guardrails that don’t just keep young people safe, but insist on youth‑directed play and real choice in AI‑shaped spaces.
Any AI system deployed at scale for young people should have to answer a simple question: Does it expand or limit their power and voice? This allows the hard, important work of becoming themselves, of imagining, trying, failing, and trying again, to stay where it belongs: with young people, playing on their own terms.
In my daughter’s group chat, nothing happens unless five girls show up, play, and make something out of nothing together. The rules are improvised, alliances shift daily, but the play is unmistakably theirs.
Katie Salen Tekinbaş is a professor of informatics at the University of California, Irvine, as well as a game designer, a member of the Connected Learning Lab, and Executive Director of Connected Camps. She is a Public Voices Fellow on Youth Well-Being and Power of The OpEd Project and Hopelab.
