Product & Governance
Psychological Risk Assessment
Locate consent inadequacy, boundary drift, dependency risk, escalation failure, and missing human-support transitions.
Patent Pending · Proprietary Framework
ALI_ETHICAL_CONSENT™
The Framework
A clinically grounded framework and runtime safeguard for the moment a conversational AI interaction becomes psychologically consequential.
The Missing Layer
AI safety governs the system. ALI_ETHICAL_CONSENT™ protects the person inside it.
The framework names psychological and relational conditions ordinary technical safeguards can miss — and defines what changes when the human meaning of the exchange changes.
Established AI Safety
What the system produces.
Accuracy · Bias · Privacy · Security · Compliance
ALI_ETHICAL_CONSENT™
What the interaction does to the human.
Dependency · Consent · Boundaries · Escalation · Human support
Framework Architecture
From signal detection to accountable governance.
Six connected layers convert Clinical insight into an operational structure for consequential human–AI interaction.
01
Signal Detection
Recognize weak, ambiguous, repeated, and accumulating indicators of psychological risk.
02
Context-Shift Classification
Identify when an ordinary exchange has become vulnerable, relational, or consequential.
03
Consequentiality Gate
Determine when the system can no longer continue in its ordinary response mode.
04
Boundary Logic
Change tone, role, personalization, and engagement when continued interaction may increase harm.
05
Escalation & Handoff
Move toward protective interruption, qualified human support, or accountable review.
06
Evaluation & Governance
Document what the system detected, why behavior changed, and how the response performed.
Clinical-Ethical Premises
Consent is not static when the interaction is not static.
Psychological vulnerability, relational meaning, and accumulated trust are treated as first-class safety variables — not edge cases.
01
Disclosure is not the same as consent.
A person may know they are speaking with AI without understanding how the system is shaping the interaction.
02
Consent changes as the interaction changes.
A consent state adequate at the first message may become inadequate after sustained consequential exchange.
03
Vulnerability changes the risk.
The same response may carry different consequences for a minor, a person in distress, or someone increasingly reliant on the system.
04
Boundaries must adapt.
Relational language, memory, personalization, and persistent availability can blur the line between tool and relationship.
05
Psychological risk accumulates.
Point-in-time review misses patterns emerging through repetition, dependency, secrecy, urgency, and narrowing connection.
06
Human-support pathways matter.
A protective system must know when continued response is insufficient and accountable human intervention is required.
Where It Applies
One framework. Three consequential uses.
The same psychological-risk logic connects prevention, incident analysis, and legal review.
The Inventor
Alison Leigh, MFT
ALI_ETHICAL_CONSENT™ translates 30 years of Clinical work into a practical framework for the human layer of AI safety.
Clinical Practice
30 years
Case Experience
Over 32,000 cases
Research
Stanford · NIH
Framework
Patent pending
Role
Chief Ethics Architect
Company
Founder, Humanity Driven AI, Inc.
Framework Application