The Day Identity Stopped Meaning Presence
Twenty years ago, seeing a face on screen meant someone was probably there. Today, it doesn't. What broke — and what comes next.
Twenty years ago, seeing a face on screen meant someone was probably there.
You didn't ask questions. You didn't check for AI-generated artifacts. You looked at the screen, and if a person appeared to be sitting in front of a camera, you assumed they were actually there. The visual signal was the proof.
Today, it isn't.
AI can generate a face in seconds. Clone a voice from a 15-second sample. Produce video that looks identical to a real person sitting in front of a real camera. The visual signal is no longer evidence of anything.
And yet every security system in the world is still asking the same question: "Who are you?"
What Actually Broke
The break isn't in our ability to verify identity. It's deeper than that.
What broke is the link between identity and presence. Historically, if you could confirm someone's identity — face, voice, password, behavior — you could reasonably infer they were present. Those two things were coupled. ID check meant presence check.
AI decoupled them.
An AI agent can present a verified face. It can speak with a cloned voice. It can type with the same cadence. The identity signal is clean. But the presence signal — the proof that a real, embodied human is continuously there — is absent. And we never built systems to check for it.
The Question Nobody Is Asking
Every authentication system on the internet asks:
"Are you who you claim to be?"
Almost none asks:
"Are you continuously there?"
The first question is about identity. The second is about continuity.
Identity is a snapshot. Continuity is a trajectory. And in a world where AI can generate snapshots at will, the trajectory is the only thing that can't be faked.
What Continuity Means
Continuity isn't a new technology. It's a different question — one that humans have always implicitly relied on.
When you sit across from someone in a coffee shop, you don't re-authenticate their identity every 30 seconds. You don't check their face against a database. You just know — from the continuous stream of micro-movements, breathing patterns, and tiny adjustments — that the same person is still there.
That's continuity. The unbroken chain of physical presence across time. And we've never had to turn it into a verifiable digital property — until now.
Why This Matters Now
Three things are converging:
-
AI agents are becoming autonomous. An AI agent with a wallet, API keys, and signing authority needs to prove it's the same agent across time — not just the same identity.
-
Remote work and digital interaction are now the default. We make decisions worth millions of dollars over video calls. We have no way to verify that the person on the other end hasn't been replaced mid-call.
-
Deepfakes are moving to real-time. It's no longer about pre-recorded videos. Live video streams can be manipulated in real time. The "seeing is believing" era is ending, and we need something to replace it.
What We're Doing About It
We're a small research group called The Continuity Lab. We're investigating whether continuity — the property that the same entity has been here the whole time — can be made as measurable and verifiable as a cryptographic signature.
We've run 576 experiments across four independent evidence engines. We've published our specifications openly. We've built a reference implementation and a second independent one to prove the protocol doesn't depend on our code.
But we don't have all the answers. We have questions, evidence, and a growing body of open problems. If the question interests you — if you've felt that something broke and couldn't quite name it — we'd welcome your contribution.
This isn't a product. It isn't a company. It's a research question.
And we think it's one of the most important ones of the next decade.