BAAI/S-1 · v0.9 Public Comment Draft
BAAI/S-1: AI Documentation Standard for Behavior Analysis
The v0.9 Public Comment Draft is open through Q2 2027. We invite practitioners, researchers, educators, and organizations to read the draft and submit feedback that will shape v1.0, ratified by founding members in 2027.
The Standards define minimum expectations for documenting AI use in behavior-analytic practice: what must be disclosed, what must be verified, how human oversight is preserved, and how records are maintained to support the highest level of clinical quality.
CONFORMANCE LEVELS:
Three cumulative levels of conformance.
L1
Disclosed
AI use is lawful, PHI-safe, disclosed, and covered by policy. The ethical baseline.
L2
Verified
Every AI-assisted note, document or product passes a defined human verification workflow before entering the record.
L3
Audit-Ready
The organization can prove L1 and L2 to a third party from records alone.
Free Resource
37 requirements. One checklist. Audit-ready documentation.
The checklist operationalizes BAAI/S-1 item by item — criteria built from the documentation deficiencies OIG auditors actually cited. Preview the first 5 items; enter your email to subscribe and receive the full checklist.
- 1.Every AI tool that touches client data has an executed BAA on file — not HIPAA-ready marketing (S5.4.1–5.4.2)
- 2.No staff member uses free or consumer chatbots with PHI — verified by policy and asked directly in supervision (S5.4.1)
- 3.Vendor's use of your data for model training is documented and disabled or excluded for PHI (S5.4.3)
- 4.AI tools sit inside normal access-control and device policies (S5.4.5)
- 5.A written policy names the approved tools, permitted document types, and permitted roles (S5.5.1)
…32 more items in the full checklist.
Submit a comment
Comments received during the public period will be reviewed and reflected in the v1.0 revision.