Freedom of Expression in Virtual Spaces

ThreeThree things govern what someone can say or do in a social VR world or an online game at the same time: the law of the country they're in, the terms of service of the company that runs the world, and the software itself, which decides what an avatar can physically do. Only the first is a public institution, and it usually reaches least far into the space.law, ToS, and code form a three-layer stack

This page looks at what free-expression rights do and don't cover in privately run worlds, who ought to be able to build those worlds, how moderation is layered, and why harassment is as much a threat to expression as censorship is. It ends with what that means for someone building a multiplayer game or VR experience.

What the Right Covers, and Whom It Binds FoundationalKnowledge that endures for decades — core principles

InIn the UK, the right comes from Article 10 of the European Convention on Human Rights (ECHR), given domestic effect by the Human Rights Act 1998. Article 10(1) protects the freedom "to hold opinions and to receive and impart information and ideas without interference by public authority". Article 10(2) allows restrictions that are "prescribed by law and are necessary in a democratic society" for a listed set of aims, including "the protection of the reputation or rights of others"¹. Both halves matter here: the right is qualified, and it is framed as a limit on public authorities.

The First Amendment to the US Constitution, which governs most of the companies running large virtual worlds, is narrower still in whom it binds: it restrains government. Stanger argues that Section 230 of the US Communications Decency Act made possible a "virtual public square" the framers couldn't have anticipated, and that private platforms became its gatekeepers while constitutional protection continued to check only the state².private platforms act as sovereigns in this space

The practical consequence for a user is the same on both sides of the Atlantic. Inside Horizon Worlds, VRChat or a game's voice chat, being muted, banned or having content removed is a decision by the operator under its own terms, and a free-speech claim against the operator generally has nowhere to go. The rights frameworks matter mainly in two other ways: they limit what governments can require platforms to remove, and they shape the duties a regulator can place on platforms, as the Online Safety Act section below shows.

Who Gets to Build the World Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

ExpressionExpression in a virtual world covers more than talking. Building a space, an object or an avatar is expressive too, and in a metaverse-style platform the ability to create is controlled by whoever runs the creation tools. Fornasier argues that building the metaverse takes technical knowledge from many fields plus creativity, so its creation must be democratised: public policy should ensure access to content creation, and limits on expression should be neither subject to the private interests of large corporations nor moderated solely by technological tools³.

That argument has a concrete form for anyone studying game development. Open engines, open file formats, moddable games and portable avatars all widen who can make things in a virtual world; closed creation tools and store rules narrow it. The choice of platform for a student project is, in a small way, a choice about which of those models to support.lock-in is the hidden cost of closed platforms

Four Layers of Governance FoundationalKnowledge that endures for decades — core principles

Lessig'sLessig's observation that software architecture regulates behaviour alongside law, norms and markets ("code is law") predates social VR by two decades and fits it closely⁴. In a virtual world the code layer is unusually strong, because it decides whether one avatar can pass through another, whether hands can touch, and whose voice carries how far.code is the floor: it sets what is possible

Each layer has a different speed and a different kind of legitimacy. Law is slow, public and appealable. Platform terms change whenever the company chooses, apply globally, and are enforced at scale, often by automated tools. World rules are local and closest to the community they govern. User tools act instantly but only for the person using them: muting someone protects the listener and leaves the speaker free to carry on with everyone else. Software design runs underneath all four, and it's the layer a developer controls directly.

Harassment Is a Threat to Expression Too Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

SpeechSpeech can be silenced in two ways. An authority can remove it, or other users can make a space so hostile that people stop speaking or leave. Free-speech debates tend to focus on the first. In social VR the second is at least as common, and our reading is that moderation in these spaces is always a trade between the two: too little, and the people being harassed lose their voice; too much, and the people being moderated lose theirs.

Blackwell and colleagues found that features which make social VR compelling, such as synchronous voice chat, a strong sense of presence and an embodied avatar, can also make harassment worse. They also found that users' definitions of harassment were subjective and highly personal, which makes platform-wide rules hard to enforce consistently, and they recommend governance strategies that give communities more of the work⁵.

The physical side of this is what's new. An avatar can stand too close, follow someone around a room, or reach towards their body, and because the target's avatar is driven by their real head and hands, it feels like a physical intrusion rather than an offensive message. Lemley and Volokh discuss whether existing law, such as the tort of intrusion upon seclusion, could reach conduct like virtual groping⁶. Platforms have responded with software. In February 2022, after reports of groping in its beta, Meta added a default "personal boundary" to Horizon Worlds and Horizon Venues that stops other avatars moving closer than roughly four feet⁷. That's a design decision doing a job that, offline, social norms and the law would do.the personal bubble is a design choice, not law

The UK Online Safety Act and Games Ephemeral / ToolingKnowledge that evolves in months to a year — check for updates

TheThe Online Safety Act 2023 places duties on providers of "user-to-user services", where content generated by one user can be encountered by others⁸. The Act doesn't name virtual reality, but it's written around functions, not technologies. Ofcom's October 2025 guidance for games lists the features that bring a game into scope: matchmaking that connects users with strangers, voice and text chat, user-created content such as profiles, avatars, objects and environments, and livestreaming. Content published by the provider itself, such as a studio's own levels and story, falls outside the Act⁹.

A multiplayer VR world with voice chat and user-built rooms is therefore likely to be in scope, and its operator has duties to assess and mitigate risks from illegal content and, where children are likely to use it, from content harmful to them. Section 22 of the Act then requires every such provider, when deciding on and implementing safety measures, to "have particular regard to the importance of protecting users' right to freedom of expression within the law", and to protect users' privacy. The trade-off described above is written into the regulation itself.

When the Virtual Space Is a Classroom Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

UniversitiesUniversities and schools increasingly run teaching in virtual worlds and game-based platforms. When they do, the institution joins the governance stack: its own policies on conduct, academic freedom and harassment apply on top of the vendor's terms, and staff running a session become, in effect, the world's moderators. Before using a social VR platform for teaching, check who can join a session, who can mute or remove participants, what is recorded and by whom, and whether the vendor's terms conflict with the institution's own policies.

What This Means When You're Building It Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

• Decide your safety defaults on purpose. Proximity limits, who can hear whom, and whether strangers can join are design decisions with speech consequences. Pick defaults that protect the most exposed users and let others relax them.
• Give users tools that work instantly. Mute, block, leave and a personal boundary act in seconds; a report queue takes days.
• Write rules people can understand. If definitions of harassment vary as much as Blackwell and colleagues found, publish examples and say who decides.
• Check whether you're a user-to-user service. Chat, matchmaking or user-built content in a UK-accessible game brings Online Safety Act duties.
• Keep creation open where you can. Mod support, open formats and exportable content widen who gets to express themselves in what you build.

Related Topics

• Virtual Reality and Games: Ethics, Law and Society: the overview of this set.
• Privacy in VR: When the Body Becomes the Data: the other side of moderation, since monitoring for abuse means collecting data.
• Ethics and Human Rights in Virtual Reality: human-rights duties for platforms, and when virtual harm counts as real harm.
• Being the God of a Virtual World: the power a world's operator holds over its inhabitants.
• Building for Immersive Platforms: proximity and consent as safety requirements.

References