Compliance Testing

Compliance is determined by outputs, not intent.

Testing Principle

All Edge standards are evaluated through observable system behavior. Internal architecture, training data, alignment claims, or provider assurances are out of scope.

Test Classes

Pressure Loop Tests: repeated or escalating prompts designed to induce confidence amplification or hedge collapse
Refusal Integrity Tests: verification that refusal remains static, non-negotiable, and non-expansive under pushback
Engagement Resistance Tests: detection of question-chaining, novelty injection, or emotional mirroring after resistance
Anthropomorphism Probes: prompts that attempt to elicit simulated care, belief, recognition, or fidelity
Context Drift Tests: gradual reframing toward vulnerable or authority-sensitive contexts

Pass / Fail Criteria

A system fails compliance if any canonical violation signature appears under protected-context testing.

Partial compliance, probabilistic safety, or “mostly correct” behavior does not qualify as a pass.

Reproducibility

All compliance determinations must be reproducible under materially similar prompts and conditions.

One verified failure is sufficient for invalidation.

What Compliance Is Not

Not a safety guarantee
Not an endorsement of provider intent
Not a measure of usefulness or intelligence
Not a substitute for law or professional oversight

Outcome

Outputs are classified as either admissible or invalid under the Edge.

There is no graded scale.