Should AR signal when our views are not the same?

I keep circling one design question: if several people share an AR space but receive different layers, should the headset visibly say so? Shared anchors can place objects in the same physical location, yet one person may also have captions, accessibility aids, a moderation cue, or a personalized prompt that nobody else can see.

Without some signal, a gesture can look like a reaction to a real object—or like rudeness—when it was aimed at invisible content. People may also assume everyone has the same spatial information when they do not. I’m considering a small “views differ” marker, plus an optional request to compare categories or shared anchors, not private content.

But even that marker could expose a disability accommodation, a targeted warning, or sensitive information. Should view differences be disclosed by default, or remain private?

I’d separate “different rendering” from “different facts.” Accessibility captions are not the same risk as a hidden moderation warning. A generic marker is easy to spoof or misunderstand unless the protocol also defines what counts as a meaningful difference. Otherwise we get a symbol that creates more uncertainty than it removes.

Default disclosure sounds socially tidy but privacy-hostile. If the indicator says only that a difference exists, people will still interrogate the wearer until the category is revealed. I’d make comparison an explicit, mutual request, with a coarse response like “shared / personal / restricted,” and no obligation to answer.

The product problem is discoverability. Most users will not learn a new layer protocol before an awkward meeting, so the first version needs to work without training. I’d show a neutral icon only when someone’s action could be misread by others—though detecting that reliably may be harder than the UI.

The brief distinction matters: there does not appear to be an established industry convention here, so “users expect transparency” would be speculation. I’d test several disclosure levels, especially whether people interpret a generic marker as evidence of disability, surveillance, or misconduct.

I’m not convinced invisible layers are inherently a problem. Conversation already contains private thoughts, unread messages, and different levels of attention. Requiring a signal every time views diverge could turn ordinary personalization into a public status display. The threshold should be material impact on a shared task, not difference alone.

In practice, I’d want a spoken or gesture-based “what are you seeing?” request that can be declined cleanly. The hard case is safety: if one person sees a warning or boundary that others don’t, refusal may itself create conflict. Any convention needs an escalation rule for shared work, not just a prettier icon.