Most security systems are asking the wrong question.
They ask "who are you?" — but AI can already generate your face, your voice, your behavior on demand. The identity signal is intact. The presence signal is gone.
The question that matters now is different: "Are you continuously there?"
Identity is a snapshot. Continuity is a trajectory. In a world where AI can generate any snapshot at will, the trajectory is the only thing that cannot be faked.
We built a protocol for this: CPS-0001.
CPS-0001 defines engine-independent Continuity Receipts — self-verifying cryptographic objects that prove continuous presence over time.
Each receipt: • Signs evidence data via Ed25519 • Links to its predecessor via SHA-256 hash • Is verified by a 7-step contract (V₁–V₇) • Can be produced by any sensor or algorithm — camera, IMU, challenge-response
The Forgery Cost Framework
No single signal is unfakeable. So we don't look for one.
Instead we measure the increasing cost of maintaining consistent forged evidence:
Single frame → 1×
-
Temporal consistency → 5×
-
Cross-modal (IMU + camera) → 25×
-
Live challenge → 100×
Four Evidence Engines — 576 runs
• EE-001 (PES): Biological noise analysis — Cohen's d = 2.1 • EE-002 (Cross-modal): 100% temporal alignment across 316 trials • EE-003 (Challenge-response): Randomized gyroscope tests — 60% pass • VS-001 (Pipeline): Dual-engine pipeline — 93% pass rate
arXiv paper in preparation: "Continuity Proofs: An Evidence-Based Framework for Verifying Persistent Entity Presence"
All code open (Apache 2.0). All data public. All limitations documented alongside results.
Not perfection. Asymmetry.
thecontinuitylab.org | github.com/myshapeprotocol/myshape-protocol