AI Usage Policy - PsyData Labs

Definitions

"AI System" means any machine learning model, AI agent, inference pipeline, or algorithmic decision system operated or provided by PDL.
"AI Output" means any result, recommendation, prediction, classification, or generated content produced by a PDL AI System.
"High-Stakes Decision" means any decision with significant consequences for a person's health, legal rights, financial wellbeing, or safety.
"Human Oversight" means review and accountability by a qualified human professional before a High-Stakes Decision is made or acted upon.
"Bias" means systematic and unfair differences in AI Outputs based on protected characteristics including race, gender, disability, or other legally protected attributes.
"Incident" means a significant AI failure, safety event, unexpected harmful output, or adverse event involving a PDL AI System.

Permitted Uses

PDL AI Systems may be used for:

Academic and scientific research within authorized scope;
Behavioral analytics within authorized deployment parameters;
Clinical decision support with mandatory qualified human oversight;
Product development, testing, and evaluation;
Educational and training applications;
Internal operational automation within PDL governance policies. 12. Prohibited Uses
Use of PDL AI Systems for any of the following is strictly PROHIBITED:

Psychological manipulation of individuals or groups without informed consent;
Deceptive AI personas designed to deceive users that they are interacting with humans;
Mass or individual surveillance without legal authorization and informed consent;
Discriminatory profiling based on protected characteristics;
Autonomous clinical diagnosis, treatment decisions, or prescribing without qualified human clinical oversight;
Autonomous High-Stakes Decisions without Human Oversight;
Adversarial attacks on other AI systems;
Military weapons development or autonomous lethal systems without explicit written authorization;
Any use that violates applicable law, including privacy law, anti-discrimination law, or AI-specific regulation. 13. Human Oversight Requirements
All High-Stakes AI Outputs must be reviewed by a qualified professional before being acted upon. PDL AI Systems are not designed or authorized to replace qualified professional judgment. Licensees and Users must implement Human Oversight procedures for any deployment in clinical, legal, financial, or safety-critical contexts.

Output Transparency

AI-generated outputs must be disclosed as AI-generated where required by applicable law or where a reasonable person would expect disclosure. PDL's AI outputs in clinical and legal contexts must be accompanied by appropriate disclaimers that outputs do not constitute professional advice absent Human Oversight.

Bias and Fairness Obligations

Users and Licensees must monitor for bias in AI Outputs within their deployment contexts. Identified bias that may cause harm or discrimination must be reported to PDL. PDL will use reports to improve model fairness over successive versions.

Incident Reporting

Significant AI failures, safety Incidents, unexpected harmful outputs, or adverse events involving PDL AI Systems must be reported to PDL within seventy-two (72) hours of discovery. Reports should include: description of the Incident, affected system and version, user population affected, observed outputs, and any remediation steps taken. Reports to: incident@psydata.org.

Policy Updates

PDL may update this AI Usage Policy as AI technology evolves, as regulatory frameworks develop, and as new risks are identified. Material updates will be communicated to Licensees with reasonable notice. Continued use of PDL AI Systems following notice of updates constitutes acceptance of the updated Policy.