Eye tracking was supposed to be the next big thing in VR about three years ago. Then it mostly just… rendered things more efficiently in the background. Foveated rendering—where the headset only draws full detail where you’re looking—is genuinely impressive engineering, but it’s invisible to users by design. You’re not supposed to notice it.

What’s changed in early 2026 is that eye tracking is starting to do things you actually interact with. And some of them are surprisingly good.

Beyond Foveated Rendering

The Quest Pro introduced eye tracking to Meta’s ecosystem back in 2022, and Apple baked it into Vision Pro from day one. But most VR applications treated it as a rendering optimisation or a novelty—look at a button to highlight it, blink to confirm. Useful in demos, rarely compelling in practice.

The shift I’m seeing now is developers treating eye gaze as a genuine input channel, not a mouse replacement but something distinct. Your eyes move differently from your hands. They’re faster, less precise, and constantly revealing where your attention sits.

Tobii, the Swedish company that’s been doing eye tracking longer than most VR companies have existed, published research showing that gaze-based interfaces reduce selection time by 30-40% compared to controller pointing in information-dense environments. Think dashboards, data visualisation, and complex menus. The gains come from the fact that your eyes arrive at the target before your hand does—by a lot.

What Gaze-First Design Looks Like

A few applications are building for eye tracking as a primary interaction, not an add-on.

visionOS applications on Apple Vision Pro are the most mature examples. Apple’s entire UI model assumes eye tracking works. You look at something, it highlights, you pinch to select. After a few days, the interaction feels natural enough that going back to a controller-based menu feels clumsy. The precision isn’t perfect—small buttons close together cause issues—but designers are adapting by spacing interactive elements further apart and using larger touch targets.

Medical imaging in VR is another area where gaze shines. Radiologists reviewing 3D scans in VR can look at a region of interest and have the interface automatically zoom and enhance that area. It’s faster than navigating with controllers, and it maps to how radiologists already work—they scan an image with their eyes, then focus on anomalies.

Training and assessment applications are using eye tracking not for input but for evaluation. A VR-based safety training module can track whether the trainee actually looked at the hazard sign, checked the emergency exit, or glanced at the pressure gauge before opening the valve. That data is more reliable than asking someone to click on things in the right order—it shows what they naturally pay attention to.

The Privacy Elephant in the Room

Eye tracking data is extraordinarily personal. Your gaze patterns reveal what catches your attention, how long you look at things, where your eyes linger. Researchers have shown that eye tracking data can indicate cognitive load, emotional state, fatigue, and even neurological conditions.

Mozilla’s research on XR privacy raised important questions about how this data should be handled. Most headset manufacturers say eye tracking data stays on-device and isn’t uploaded to servers. But application developers have access to gaze data through APIs, and the policies governing what they do with it are inconsistent at best.

This isn’t theoretical. An advertising platform with access to eye tracking data knows not just that you looked at a virtual billboard, but how long you looked, whether your pupils dilated (indicating interest), and whether you looked back at it. That’s a qualitatively different kind of attention data than “this ad was displayed on screen.”

The VR industry needs to get ahead of this before regulators do it for them. Opt-in consent, data minimisation, and clear retention policies aren’t optional when you’re tracking someone’s literal gaze.

Accessibility Gains

One genuinely positive application: eye tracking is making VR accessible to people with limited mobility.

If you can’t hold controllers or make hand gestures, a gaze-and-dwell interface—look at something for a set duration to select it—provides a way into VR experiences that was previously unavailable. It’s not fast, and it requires careful UI design to avoid accidental selections, but it works.

Several rehabilitation facilities in Australia are experimenting with gaze-controlled VR for patients recovering from spinal injuries or strokes. The patient can explore a virtual environment, select items, and complete therapeutic tasks using only their eyes. The engagement levels are higher than traditional screen-based rehab exercises, apparently because the sense of presence and agency is stronger.

What Needs to Improve

Current eye tracking in consumer headsets isn’t reliable enough for every user. Glasses wearers often get worse calibration. Different eye shapes and colours affect tracking accuracy. And calibration drift—where the tracking gradually becomes less accurate during a session—is a known issue that manufacturers are still working on.

The latency between where you look and the interface responding needs to be under 20 milliseconds to feel natural. Most current headsets hit that target most of the time, but not always. When it’s off, the experience feels laggy and disconnected in a way that controller-based interaction doesn’t.

Developers also need better design patterns. Building a gaze-first interface isn’t the same as building a touch or controller interface, and the UX conventions are still being established. Apple has the most coherent design language for this, but it’s locked to their ecosystem.

Eye tracking in VR is past the “cool demo” stage and into the “solving real problems” stage. The next two years will determine whether it becomes a standard interaction model or remains a secondary feature. Based on what I’m seeing, I’d bet on the former.