IAICA Discussion Framework: Ethical AI Care Standard.
A public discussion and risk-assessment draft (v0.1) evaluating ambient monitoring, conversational AI, and sensor tools in vulnerable population and caregiver environments.
IAICA Ethical Care Discussion Framework
Current Status: Public Discussion Draft v0.1 (RFC 2026-A)
Automated caregiving technologies and ambient monitors operating around vulnerable individuals cannot be audited solely via benchmark code tests. Our draft framework assesses 6 proposed dimensions:
Six Proposed Evaluation Dimensions
1. Cognitive Dignity & Anti-Objectification
Prohibits deceptive emotional simulation that exploits cognitive decline; mandates transparent AI identity and respectful interaction.
2. Sensory Risk Assessment
Audits acoustic alerts, lighting transitions, and motion frequencies to evaluate sensory burden and prevent alarm fatigue without claiming zero harm.
3. Caregiver Autonomy & Burden Evaluation
Evaluates task burden, usability, and caregiver feedback to assess whether systems alleviate stress rather than adding administrative data chores.
4. Contextual Boundary & Edge Privacy
Evaluates minimal data collection, local processing feasibility, user consent, and transmission safeguards in intimate living spaces.
5. Identifiable Human Override & Escalation Procedures
Proposed criterion: assess whether users and caregivers can identify, access, and test human override and escalation procedures, including their limitations and foreseeable failure modes.
6. Community & Vulnerable Group Calibration
Algorithms must demonstrate responsiveness to real feedback gathered from community care circles and vulnerable population advocates.
Institutional Scope & Legal Distinction
• Independent Public Discussion Draft: Version 0.1 represents a non-profit risk-assessment discussion draft. It does not constitute statutory medical device approval (e.g., FDA 510(k), CE-MDR).
• Strict Separation from Training Verification: The IAICA Discussion Draft applies exclusively to software and hardware risk evaluations. It is completely separate from individual community workshop completion records (accessible at /verify).