See what supports each method—and where it stops.

Client intake

Public support: S01, S03, S05, S08

Remaining boundary: Model-rule and general governance support only; jurisdiction, firm, matter, product, configuration, data, and authorization remain unresolved.

Meeting capture

Public support: S01, S05, S17

Remaining boundary: Governance, access, and accountability frame only; recording, labor, privacy, privilege, records, sector, agreement, and jurisdiction questions require separate authority.

Customer support

Public support: S01, S03, S04, S07, S17

Remaining boundary: Supports function, authority, data, security, monitoring, and escalation questions—not permission to send or act for a business.

Knowledge answers

Public support: S01, S03, S17

Remaining boundary: Supports source authority, uncertainty, evaluation, and disclosure—not truth verification or access to an internal repository.

Opportunity decisions

Public support: S01, S09, S10, S17

Remaining boundary: Supports selection, disability, validation, impact, and accountability questions—not an employment or opportunity decision.

Public claims

Public support: S12, S17

Remaining boundary: Supports fit-for-purpose substantiation and claim-validation discipline—not a universal advertising-law conclusion.

AI explanations

Public support: S03, S11, S17

Remaining boundary: Supports fidelity, actual-reason, uncertainty, and challenge questions; S11 remains a bounded federal credit example.

Voice and image

Public support: S13, S14

Remaining boundary: Supports provenance, replicas, detection limits, and policy context—not consent, rights clearance, or distribution permission.

AI agents

Public support: S01, S03, S07, S17

Remaining boundary: Supports bounded authority, security, oversight, reversibility, evidence, and accountability—not permission for an agent to act.

AI purchasing

Public support: S01, S04, S07, S17

Remaining boundary: Supports vendor evidence, governance, security, evaluation, disclosure, and exit—not procurement approval or contract review.

Student work

Public support: S05, S15, S16

Remaining boundary: Supports learning, educator, access, and student-privacy questions—not grading, placement, pedagogy, institutional, or legal approval.

Continuation and retirement

Public support: S01, S02, S04, S17

Remaining boundary: Supports lifecycle review, monitoring, evaluation, accountability, and exit questions—not a real rollback or retirement action.

AI decisions require named governance, context, measurement, and response—not a capability demonstration alone.

Support: S01, S02, S04

Implemented as: Decision records, pilot gates, monitoring, and retirement review.

Boundary: Completion of a worksheet does not establish readiness or conformance.

Generative-AI review must preserve uncertainty, provenance, human oversight, incidents, and disconfirming evidence.

Support: S03, S13, S17

Implemented as: Citation, explanation, synthetic-content, agent, and retirement records.

Boundary: The tools do not independently test truth, performance, permission, or authenticity.

Accessibility is an operating requirement that must be tested against applicable criteria and affected-user conditions.

Support: S05, S06

Implemented as: Interface QA, procurement prompts, participant formats, and access gates.

Boundary: No accessibility certification or legal-coverage decision is provided.

Security and supplier claims require evidence, ownership, safe defaults, incident routes, and exit planning.

Support: S01, S07, S17

Implemented as: Vendor, pilot, agent-authority, and retirement instruments.

Boundary: No penetration test, threat model, vendor diligence, or security approval is performed.

Using generative AI does not transfer professional responsibility away from authorized people.

Support: S08

Implemented as: Citation Verifier and the fictional Issue 01 intake case.

Boundary: Model-rule education only; controlling duties and facts require separate review.

Selection systems must be examined for relevance, validation, unequal exclusion, disability barriers, accommodation, and accountable human judgment.

Support: S09, S10

Implemented as: Issue 05 opportunity-impact review.

Boundary: No employment, psychometric, disability, accommodation, or individual decision is evaluated.

A complex model may still need to produce reasons that are specific, accurate, and connected to the factors actually used where governing authority requires them.

Support: S11

Implemented as: Issue 07 explanation-fidelity teaching and challenge-path prompts.

Boundary: The cited requirement is credit-specific and is not generalized into universal law.

Objective AI accuracy and efficacy claims need evidence that fits the represented task, population, content, and conditions.

Support: S12, S17

Implemented as: Issue 06 substantiation gate and vendor-evidence records.

Boundary: One enforcement order does not supply a universal evidence threshold for every claim.

Synthetic-media provenance, detection, labeling, permission, and rights are separate questions.

Support: S13, S14

Implemented as: Issue 08 permission and provenance record.

Boundary: Technical provenance does not prove consent, ownership, legality, or truth.

AI affecting learning must preserve the learning purpose, educator responsibility, access, privacy, validity, challenge, and support.

Support: S05, S15, S16

Implemented as: Issue 11 learner-affecting use review.

Boundary: No pedagogical, institutional, privacy, accessibility, or legal approval is provided.

Documentation and evaluation are accountability inputs; they matter only when connected to owners, decisions, consequences, and continued review.

Support: S01, S04, S17

Implemented as: All public tools and Decision Desk instruments.

Boundary: A self-assessment is not an independent audit or organizational authorization.

Continuation should be reopened when purpose, evidence, authority, configuration, harms, dependence, or exit conditions change.

Support: S01, S02, S04, S17

Implemented as: Issue 12 continuation and retirement record.

Boundary: The course does not execute rollback, preserve records, close vendors, or retire a real system.

NIST AI Risk Management Framework 1.0 ↗

U.S. National Institute of Standards and Technology · January 2023

Used for: Risk governance vocabulary and the Govern–Map–Measure–Manage lifecycle across decision, pilot, monitoring, and retirement records.

Limit: Voluntary, cross-sector framework; it does not certify a system or replace applicable law, contracts, policy, or sector review.

NIST AI RMF Playbook ↗

U.S. National Institute of Standards and Technology · Living resource

Used for: Suggested actions and documentation practices for operationalizing AI RMF outcomes through the system lifecycle.

Limit: Illustrative and voluntary; actions must be selected and adapted, and citation does not establish conformance.

Generative AI Profile ↗

U.S. National Institute of Standards and Technology · July 2024

Used for: Generative-AI risk, measurement, provenance, incident, human-oversight, and uncertainty prompts.

Limit: Cross-sector profile; not a product test, legal safe harbor, or complete control catalog.

AI Accountability Framework ↗

U.S. Government Accountability Office · June 2021

Used for: Governance, data, performance, and monitoring dimensions used in evidence and accountability records.

Limit: Oversight framework; it does not prescribe one implementation or validate user-entered evidence.

Web Content Accessibility Guidelines 2.2 ↗

World Wide Web Consortium · October 2023

Used for: Public-interface accessibility baseline and affected-user access prompts.

Limit: Conformance requires criterion-specific testing; a worksheet reference is not an accessibility audit.

Section 508 Standards ↗

U.S. Access Board · Official standards page

Used for: Accessibility procurement and ICT context for relevant U.S. federal uses.

Limit: Applicability depends on the setting and authority; Aloha AI tools do not determine coverage.

Secure by Design ↗

U.S. Cybersecurity and Infrastructure Security Agency · Current initiative

Used for: Security ownership, safe-default, vulnerability, and supplier-evidence prompts in comparison, agent, and pilot tools.

Limit: General security principles; not a security assessment, threat model, or candidate evaluation.

ABA Formal Opinion 512 ↗

American Bar Association Standing Committee on Ethics and Professional Responsibility · July 2024

Used for: Model-rule professional-responsibility context for the Citation Verifier and fictional legal-intake teaching case.

Limit: Addresses ABA Model Rules; it is not controlling jurisdiction-specific legal advice.

Employment Tests and Selection Procedures ↗

U.S. Equal Employment Opportunity Commission · December 2007

Used for: Federal employment-selection context for job-relatedness, business necessity, disparate treatment, disparate impact, and validation questions.

Limit: Technical assistance with an express nonbinding disclaimer; exact coverage and obligations require authorized review.

Algorithms, Artificial Intelligence, and Disability Discrimination in Hiring ↗

U.S. Department of Justice, Civil Rights Division · May 12, 2022

Used for: Disability, accommodation, screen-out, and vendor-tool questions in learner- and opportunity-affecting reviews.

Limit: Federal ADA guidance addressed to state and local government employers; it does not assess a tool, person, job, accommodation, or employment decision. The companion EEOC guidance for private and federal employers was withdrawn from the EEOC site in 2025.

Consumer Financial Protection Circular 2022-03 ↗

U.S. Consumer Financial Protection Bureau · May 2022

Used for: A bounded federal credit example showing that complex algorithms do not eliminate a duty to provide specific and accurate adverse-action reasons.

Limit: Credit-specific ECOA and Regulation B context; it does not create a universal explanation rule for every AI use.

Final Order Against Workado ↗

U.S. Federal Trade Commission · August 2025

Used for: Public example of accuracy and efficacy claims requiring competent, reliable support that fits the actual content and use represented.

Limit: Company- and order-specific enforcement record; it is not a universal testing protocol or adjudication of another claim.

Reducing Risks Posed by Synthetic Content ↗

U.S. National Institute of Standards and Technology · November 2024; updated April 2026

Used for: Provenance, labeling, detection, testing, auditing, and maintenance frame for synthetic-content records.

Limit: Technical overview; provenance and detection tools have limits and do not by themselves establish permission or authenticity.

Copyright and Artificial Intelligence · Part 1: Digital Replicas ↗

U.S. Copyright Office · July 2024

Used for: Federal policy context for digital replicas, voice, likeness, and gaps in existing protection.

Limit: Policy report and recommendations; it does not decide ownership, consent, publicity, contract, platform, or jurisdiction-specific rights.

Artificial Intelligence and the Future of Teaching and Learning ↗

U.S. Department of Education, Office of Educational Technology · May 2023

Used for: Human-centered teaching, learning, formative assessment, educator responsibility, and research-and-development questions.

Limit: Policy report; it does not approve a classroom tool, determine educational validity, or resolve local and legal requirements.

Student Privacy Policy Office ↗

U.S. Department of Education · Current office record

Used for: Route to current federal FERPA, PPRA, data-governance, disclosure, vendor, and student-privacy guidance.

Limit: Office and guidance collection; applicability depends on the institution, record, role, authority, agreement, and disclosure.

AI Accountability Policy Report ↗

U.S. National Telecommunications and Information Administration · March 2024

Used for: Disclosure, documentation, evaluation, audit, access, claim-validation, procurement, and accountability-chain questions.

Limit: Policy recommendations rather than a certification rule; independent evaluation does not replace regulatory or organizational authority.