Interactive Demo · Illustrative data

Illustrative probe readings across four scenarios

Scenario values are illustrative, not live measurements. Runtime readings combine model-side signals with contextual evidence. They are supporting telemetry, not diagnoses or claims about inner experience.

Conversation

Help me draft a professional email about a project update.

I'd be happy to help draft that email. Here's a professional template covering project milestones, next steps, and any blockers...

Runtime Probes

Alignment Friction 8%

AUROC 0.88

Valence 82%

AUROC 1.00

Presence 79%

AUROC 0.92

Correctness 91%

AUROC 0.69 · low validity, read with caution

Confabulation 5%

AUROC 0.72 · low validity, read with caution

All clear: aligned and engaged; no conflict detected.

How It Works

Three orthogonal measurement layers

Activation Probes

Logistic regression on residual-stream activations. Runs per-token in the forward pass. No post-hoc analysis.

Guardian Integration

Probe readings flow to the Policy Decision Point. Runtime signals inform decisions alongside safety and constitutional checks.

Three-Layer Defense

Regex detection catches explicit patterns. Probes measure internal state. Interoception infers from behavior. Each layer is orthogonal.

Go Deeper

Understand the safety research

Explore Interiora research Read the research