The research, made testable.

Domestic violence intake workers make life-safety triage decisions with no AI decision support designed for their setting. The tools that exist were built for adjacent contexts, have not been evaluated against trauma-informed standards, and in documented cases have caused harm through misclassification. The cost of that gap is measured in survivor outcomes.

A four-person intern cohort spent three months building a trauma-informed AI framework for DV intake settings, with privacy and safety as primary design constraints from day one. A lived-experience advisor shaped the design criteria and holds formal authorship. Language conventions followed the National Network to End Domestic Violence's style guidance throughout.

This is the first completed instrument in TIAIL's pipeline and a direct output of the flagship scoping review. A second cohort is continuing the iteration.

Standard AI bias audits measure statistical parity across demographic groups. That is necessary. It is not sufficient for systems operating in crisis settings, where harm can occur through interactions that are statistically neutral but structurally unsafe for someone in acute distress or with prior trauma with institutional systems.

This rubric translates the SAMHSA six principles of trauma-informed care into structured audit criteria for AI systems in health and social services. Designed for procurement officers, program directors, and clinical informaticists, not only researchers. Each criterion includes a clinical literacy note explaining the SAMHSA principle it derives from and why it matters operationally.

A note on who should use this: the rubric is only as honest as the person administering it. We built it for people who are willing to find problems, not confirm compliance.

Open instrument. No intake required. Anyone with a system to audit can use it.

AI-assisted risk assessment tools are being used in domestic violence intake settings today. Most were not designed for that context. None have been evaluated against trauma-informed standards. Procurement decisions are being made on the basis of vendor claims, not empirical evidence.

This framework is the primary structured output of TIAIL's flagship scoping review. It provides evaluation criteria for AI risk assessment tools in DV settings, grounded in the literature the scoping review has mapped, calibrated to the SAMHSA six principles, and designed for program directors, clinical supervisors, and funders evaluating vendor proposals now.

Current status: pre-validation draft released for practitioner and researcher critique. Final version will incorporate full scoping review synthesis and lived-experience advisor input.

Research that aggregates survivor case material without trauma-informed conduct standards is not methodologically neutral. It exposes contributors to vicarious trauma, reproduces extractive research dynamics with communities, and compromises the validity of findings by treating the research process itself as outside the scope of ethical design.

TIAIL built the infrastructure we needed and could not find elsewhere. Contents: contributor briefing protocols for researchers handling sensitive source material, opt-out provisions at every stage, debriefing access pathways, survivor-centered language conventions based on NNEDV guidance, lived-experience advisor integration with formal compensation and authorship recognition templates, and a five-type review taxonomy governing all TIAIL research output.

We are publishing it because the field needs it. Any research team working on AI in health and social services can use it, adapt it, and build on it.

Open instrument. No intake required.

Procurement criteria, deployment restrictions, and compliance requirements for AI in health and social services are being written now. What gets defined in this window will govern these systems for a decade.

This tracker monitors active federal and state legislation, agency guidance, and international frameworks relevant to trauma-informed AI governance, annotated for relevance to nonprofit and public agency operators, not only legal teams. Current coverage: 14 active federal frameworks, 6 state-level initiatives, 3 international governance instruments. Updated quarterly.