regulated-ai-governance
Find published releases on PyPI and changes in the changelog.
Governance framework for AI agents and RAG systems in regulated environments — 41 examples, 25 jurisdictions, 10 AI frameworks, 2,631 tests.
Policy enforcement, audit logging, and compliance orchestration for AI systems that must satisfy FERPA, HIPAA, GDPR, EU AI Act, APPI, MAS FEAT, and 15+ additional regulatory frameworks.
Featured Article
MACF: The 6-Component Framework Every Enterprise Multi-Agent AI System Needs
A complete architectural walkthrough of the Multi-Agent Collaborative Framework (MACF), showing how regulated-ai-governance powers the ComplianceGate and PrivacyFilter components — with working Python examples, Mermaid architecture and sequence diagrams, and a full end-to-end request trace across all six framework components.
The problem this solves
AI agents in regulated environments can access and process data they are not authorized to see. Standard agent frameworks — CrewAI, AutoGen, LangChain, Semantic Kernel, Haystack — have no concept of regulated industry access control. When you deploy an agent in healthcare, financial services, or a government agency, the framework will not tell you whether the agent is allowed to access a record type, which decisions require human review, or how to produce the audit record the regulator requires.
This library provides those answers as composable Python filters: filters evaluate encoded rules against caller-supplied context. Context values are assertions until independently verified. Configure an audit sink to retain decisions; a permitted result does not establish legal compliance. See control evidence and audit guarantees.
Architecture
User / AI Agent Request
│
▼
GovernanceOrchestrator.evaluate(context, document)
│
├── Filter 1: Identity / Data Protection Gate ──────────────► FilterResult
│ (FERPA, HIPAA, GDPR, APPI, PDPA...)
│
├── Filter 2: Responsible AI Principles Gate ───────────────► FilterResult
│ (METI, PDPC Model AI Framework, EU AI Act...)
│
├── Filter 3: Sector-Specific Gate ─────────────────────────► FilterResult
│ (MAS FEAT, MHLW, DORA, NIST AI RMF...)
│
└── Filter 4: Jurisdiction-Specific Gate ───────────────────► FilterResult
(Cabinet Office, AHRC, DTA ADM, DSIT...)
│
▼
GovernanceReport
overall_decision: APPROVED / DENIED / REQUIRES_HUMAN_REVIEW / REDACTED
is_compliant: bool
compliance_summary: str (human-readable for audit file)
Every filter is independently testable. Adding a new regulation means adding a new filter class — no existing filter is modified. Each context object is a @dataclass(frozen=True) — no mutable state passes through the governance pipeline.
Installation
pip install 'regulated-ai-governance>=0.47.1'
With AI framework extras:
pip install 'regulated-ai-governance[langchain]>=0.47.1'
pip install 'regulated-ai-governance[llama-index]>=0.47.1'
pip install 'regulated-ai-governance[semantic-kernel]>=0.47.1'
pip install 'regulated-ai-governance[haystack]>=0.47.1'
The all installation requires Python 3.11+ and excludes security-quarantined CrewAI and legacy AutoGen. Use the release floor to prevent backtracking to an older package with different support/security contracts:
pip install 'regulated-ai-governance[all]>=0.47.1'
Python support: core Python 3.10+; llama-index, semantic-kernel and all extras require Python 3.11+ for patched dependencies and reject Python 3.10 resolution. See migration notes for the installation matrix.
Runtime qualification: core Python 3.10 compatibility is separate from security maintenance and optional SDK coverage. For production, use a security-maintained Python release at its latest patch level (3.11+) within the qualified SDK matrix. Native SDK jobs qualify Python 3.11/3.12 separately from newer-interpreter core tests; see CI configuration.
Boundary upgrade: omitted/None/empty category grants deny private categories; intentionally unrestricted access requires unrestricted_categories=True. Review migration guidance before loading legacy pipelines or upgrading caller policy mappings.
Quick start: an installed core guard
This example uses only the installed package; no framework SDK, API key or repository checkout is needed.
from regulated_ai_governance import ActionPolicy, GovernedActionGuard
records = []
executed = []
guard = GovernedActionGuard(
policy=ActionPolicy(allowed_actions={"read_record"}),
regulation="application-policy",
actor_id="authenticated-user", # Supply from your verified session.
audit_sink=records.append,
require_audit=True,
raise_on_deny=True,
)
assert guard.guard("read_record", lambda: "synthetic record") == "synthetic record"
try:
guard.guard("export_records", lambda: executed.append("export"))
except PermissionError:
pass
else:
raise AssertionError("Unlisted action was permitted")
assert executed == []
assert len(records) == 2 # One decision record for each attempted action.
Your application must send every protected action through the guard and authenticate the actor. The in-memory list illustrates audit delivery; production retention needs an appropriate durable sink. A policy decision or regulation label does not establish legal compliance. See 0.47.0 release notes and audit guarantees for acknowledgment, cancellation markers and failure semantics.
For supported framework integration, install the corresponding extra. CrewAI and legacy AutoGen integrations are security quarantined: their extras are removed because their SDK graphs include outstanding ChromaDB / DiskCache advisories. Existing adapter APIs remain for compatibility, but native CrewAI support is suspended. See security policy. The CrewAI source example uses a mock tool and can patch an absent SDK; it is an illustration, not proof of a native CrewAI agent run. Native SDK checks are tracked in CI. Sector patterns such as Singapore governance are repository examples, not modules under regulated_ai_governance.examples; follow each example's configuration and dependency requirements.
Example catalog — 41 governance examples
| # | File | Jurisdiction / Domain | Frameworks Enforced |
|---|---|---|---|
| 01 | 01_basic_policy_enforcement.py |
Cross-industry | FERPA, HIPAA, GDPR, GLBA |
| 02 | 02_crewai_guard.py |
Multi-agent | CrewAI EnterpriseActionGuard |
| 03 | 03_multi_framework_orchestration.py |
Multi-framework | LangChain, CrewAI, AutoGen orchestration |
| 04 | 04_governance_audit_skill.py |
Audit | Audit record assembly + compliance reporting |
| 05 | 05_healthcare_ai_governance.py |
Healthcare | HIPAA, FDA AI/ML Action Plan, ONC |
| 06 | 06_manufacturing_ot_governance.py |
Manufacturing | IEC 62443 OT, ISA-95, safety integrity levels |
| 07 | 07_government_ai_governance.py |
U.S. Government | NIST AI RMF, Executive Order 14110, FedRAMP |
| 08 | 08_insurance_ai_governance.py |
Insurance | NAIC Model Law, FCRA §615, state insurance |
| 09 | 09_finra_sec_governance.py |
U.S. Financial | FINRA, SEC, Reg BI, CCAR |
| 10 | 10_eu_ai_act_governance.py |
European Union | EU AI Act Articles 9/10/12, Annex III |
| 11 | 11_automotive_ai_governance.py |
Automotive | ISO 26262, SOTIF, UN-R155, GDPR |
| 12 | 12_medtech_ai_governance.py |
MedTech | EU MDR, FDA 510(k), IEC 62304, HIPAA |
| 13 | 13_financial_ai_governance.py |
EU Financial | MiFID II, PSD2, DORA Article 9/28 |
| 14 | 14_insurance_ai_governance.py |
Insurance (extended) | Solvency II, ORSA, GDPR |
| 15 | 15_public_sector_ai_governance.py |
Public Sector | NIST AI RMF, ISO 42001:2023, EO 14110 |
| 16 | 16_eu_ai_act_governance.py |
EU AI Act (extended) | Full risk classification + conformity assessment |
| 17 | 17_uk_ai_governance.py |
United Kingdom | UK GDPR Art.22, ICO AI Auditing, Equality Act, DSIT |
| 18 | 18_canada_ai_governance.py |
Canada | AIDA Bill C-27, CPPA, OPC Guidelines, Québec Law 25 |
| 19 | 19_australia_ai_governance.py |
Australia | DIIS 8 AI Ethics, Privacy Act APPs, DTA ADM, AHRC |
| 20 | 20_singapore_ai_governance.py |
Singapore | PDPC Model AI Framework v2, PDPA, MAS FEAT, IMDA |
| 21 | 21_japan_ai_governance.py |
Japan | APPI 2022, METI AI Principles v1.1, MHLW, Cabinet Office |
| 22 | 22_brazil_ai_governance.py |
Brazil | LGPD Law 13.709/2018 (Art. 7/11/33/37), Brazilian AI Bill PL 2338/2023 (Art. 3/14/16/22), ANPD Guidelines 2023, CFM 2299/2021, BCB Circular 3.978/2020, CLT + MPT |
| 23 | 23_india_ai_governance.py |
India | DPDP Act 2023 (§4/6/9/12/16), MEITY AI Advisory 2024 (synthetic media + election safeguards + bias testing), IT Act §43A + IT Rules 2011, RBI/IRDAI/CDSCO sectoral AI guidance |
| 24 | 24_south_korea_ai_governance.py |
South Korea | Korea AI Framework Act 2024 (Art. 6/10 high-impact + prohibited AI), PIPA (Art. 15/23/28-2/28-3/39-3 automated decisions + profiling + cross-border), Credit Information Act Art. 20, MFDS SaMD approval |
| 25 | 25_middle_east_ai_governance.py |
Middle East (GCC) | UAE PDPL (Federal Decree-Law No. 45/2021 Art. 7/10/16) + UAE AI Ethics 2019, Saudi PDPL (Royal Decree M/19 Art. 4/NDMO) + SAMA/SFDA, Qatar PDPA (Law No. 13/2016 Art. 8/9) + Qatar AI Strategy 2030, GCCCrossBorderFilter (adequate: AE/SA/QA/BH/KW/OM) |
| 26 | 26_africa_ai_governance.py |
Africa (Kenya, Nigeria, South Africa) | Kenya DPA 2019 (§25/§30/§31 automated decisions + sensitive data + profiling), Nigeria NDPA 2023 (§25/§34) + NITDA AI Policy §3.2, South Africa POPIA 2013 (§26/§71) + FSCA AI Guidance 2023, AU cross-border adequacy framework (KE/NG/ZA/GH/MA/TN/EG) |
| 27 | 27_iso42001_compliance.py |
ISO 42001:2023 AI Management System | ISO 42001 Clause 5 (AI Policy — top management leadership), Clause 6 (Risk assessment + Annex B impact assessment), Clause 8 (Operations — human oversight + data governance + third-party AI), Clause 9 (Performance evaluation — audit trail + documented information + incident process); conformity_level: FULL/PARTIAL/NON_CONFORMING |
| 28 | 28_nordic_ai_governance.py |
Nordic/Scandinavia (Sweden, Denmark, Finland) | Sweden IMY AI Guidelines 2023 + SFS 2018:218, Denmark Datatilsynet AI Guidance 2023 + Act No. 502/2018, Finland TSV AI Guidelines 2023 + Data Protection Act 1050/2018, NordicCrossBorderFilter (intra-EEA adequate; non-EEA jurisdiction-specific GDPR Art. 46 denials) |
| 29 | 29_eastern_europe_ai_governance.py |
Eastern Europe (Poland, Czech Republic, Hungary) | Poland UODO AI Guidelines 2023 + GDPR Act Dz.U. 2018, Czech ÚOOÚ AI Guidance 2023 + Act 110/2019, Hungary NAIH AI Guidelines 2023 + Privacy Act CXII/2011, EasternEuropeCrossBorderFilter (30-jurisdiction EEA adequate set + SCC/BCR fallback) |
| 30 | 30_us_state_ai_laws.py |
US State AI Laws (Colorado, Illinois, Virginia) | Colorado AI Act 2024 SB 24-205 §6-1-1702 (high-risk AI impact assessment + human oversight + bias testing), Illinois BIPA 740 ILCS 14/15 (biometric written consent + video interview AI + third-party sharing), Virginia CDPA AI provisions Va. Code §59.1-577/578/579/581 (sensitive data + automated profiling + high-risk AI), USStateAICrossBorderFilter (IL biometric / CO high-risk / CA CPRA / VA-TX-CT opt-out matrix) |
| 31 | 31_eu_ai_act_high_risk.py |
EU AI Act 2024 High-Risk Systems | EUAIActHighRiskCategoryFilter (Art. 5 prohibited + Annex III §1-8 conformity assessment + Art. 27 FRIA + Art. 49/71 database registration), EUAIActTechnicalRequirementsFilter (Arts. 9-15 risk mgmt/data governance/technical docs/human oversight/accuracy), EUAIActTransparencyFilter (Arts. 50/13 AI disclosure/deepfake labeling/emotion recognition/instructions), EUAIActCrossBorderFilter (Art. 2 export + Arts. 51-55 GPAI technical docs/systemic risk/copyright), 8 ecosystem wrappers |
| 32 | 32_gdpr_ai_accountability.py |
GDPR AI Accountability (EU/EEA) | GDPRAutomatedDecisionFilter (Art. 22(1) legal basis + Art. 22(3) explanation right + Art. 22(4) special category data + Art. 21(2) direct marketing objection), GDPRTransparencyFilter (Art. 13(1) privacy notice + Art. 13(2)(f) logic disclosure + Art. 13(1)(f) transfer safeguards + Art. 13(2)(a) retention period), GDPRDPIAFilter (Art. 35(1) high-risk DPIA + Art. 35(3)(c) systematic monitoring + Art. 35(3)(b) special category scale + Art. 36(1) SA prior consultation), GDPRDataMinimisationFilter (Art. 5(1)(c) minimisation + Art. 5(1)(b)/89 purpose limitation + Art. 25(1) privacy by design + Art. 5(1)(e) storage limitation), 8 ecosystem wrappers |
| 33 | 33_singapore_ai_governance.py |
Singapore (PDPA + MAS FEAT + AI Verify) | SingaporePDPAFilter (§13 consent + §15A sensitive data NRIC/FIN/health/biometric + §26 cross-border adequacy AU/CA/DE/JP/NZ/UK/EU + AI Advisory §4.2 automated decision), MASFEATFilter (Fairness §2.1 + Accountability §4.1 + Transparency §5.2 + Ethics §3.3), AIVerifySingaporeFilter (§3.1 self-assessment 11 principles + §4.2 explainability LIME/SHAP + IMDA GenAI §5.1 + §4.1 fairness testing), SingaporeCrossBorderFilter (MAS TRM §4.1 financial safeguards + PDPC TIA for CN/RU/KP + MAS Cloud AWS/GCP/Azure SG whitelist + FAA-N18 FATF), 8 ecosystem wrappers |
| 34 | 34_japan_ai_governance.py |
Japan (APPI 2022 + FSA + METI AI Governance) | JapanAPPIFilter (Arts. 15/17 consent + Art. 20 specially considered PI race/creed/medical/criminal/disability + Art. 28 cross-border adequacy EU/UK + PPC AI Guidelines §3.2 automated profiling), JapanFSAAIFilter (FIEA Art. 40 suitability + FSA Principle 3 credit explainability + Insurance Business Act Art. 113 actuarial opinion + FSA stress testing), JapanAIGovernanceFilter (METI Guideline v1.1 §4 self-assessment + AI Strategy §2.1 human oversight + METI GenAI §3 eight principles + Cabinet Office §3.3 public services), JapanCrossBorderFilter (APPI Art. 28 CN/RU/KP restricted + AML FATF non-compliant jurisdictions + FSA Cloud aws_tokyo/gcp_tokyo/azure_japan_east whitelist + US adequacy pending), 8 ecosystem wrappers |
| 35 | 35_south_korea_ai_governance.py |
South Korea (PIPA 2023 + FSC + AI Basic Act 2024) | KoreaPIPAFilter (Art. 15 consent + Art. 23 sensitive ideology/beliefs/union/political/health/sexual/biometric/criminal/genetic + Art. 28-8 cross-border adequacy EU/UK/CA + Art. 37-2 automated decision explanation 2023 amendment), KoreaFSCAIFilter (FSCMA Art. 7 robo-advisor registration + CB Act Art. 26 credit scoring FSC validation + IBA Art. 176 actuarial cert + FSCMA Art. 63 algorithmic trading), KoreaAIBasicActFilter (Art. 47 high-impact impact assessment + Art. 35 transparency + Art. 36 GenAI watermark + Art. 46 critical infrastructure human oversight), KoreaCrossBorderFilter (PIPA Art. 28-8 CN/RU/KP + FSC EFTA Art. 21-2 financial safeguards + FSC Cloud AWS/GCP/Azure Seoul whitelist + PIPC biometric notification), 8 ecosystem wrappers |
| 36 | 36_brazil_lgpd_ai_governance.py |
Brazil (LGPD + ANPD + Sectoral AI) | BrazilLGPDFilter (LGPD Arts. 7/11/20/33: legal basis + sensitive explicit consent + cross-border adequacy CN/RU/KP non-adequate + automated decision human review), ANPDAIFilter (LGPD Arts. 9(V)/10§3/38 + ANPD Resolutions: DPIA for high-risk + AI profiling transparency + legitimate interest LIA + credit scoring SERASA/SCR), BrazilSectoralAIFilter (CMN 4.993/2022 BCB notification + CFM 2.314/2022 health guidelines + ANATEL 740/2020 telecom framework + TSE 23.732/2024 electoral deepfake labeling), BrazilCrossBorderFilter (LGPD Art. 33 + AML Coaf 36/2021 + FATF Black List jurisdictions + ANPD biometric authorization), 8 ecosystem wrappers |
| 37 | 37_india_dpdp_ai_governance.py |
India (DPDP Act 2023 + MeitY + Sectoral AI) | IndiaDPDPFilter (DPDP Act 2023 §5 notice + §6 consent + §9 sensitive data financial/health/biometric/children + §16 cross-border RU/CN/KP/IR non-adequate + §13 grievance redressal), MeitYAIFilter (Draft §4.2 high-risk impact assessment + Advisory §3.1 explainability + March 2024 §2 GenAI labeling + Draft §5.3 human oversight), IndiaSectoralAIFilter (RBI/2023-24/73 financial AI + ICMR healthcare ethics + TRAI telecom consent + IRDAI insurance circular), IndiaCrossBorderFilter (DPDP §16 restricted countries + MeitY cloud empanelment aws_mumbai/gcp_mumbai/azure_india_central + sensitive training data export + RBI/2021-22/57 payment data), 8 ecosystem wrappers |
| 38 | 38_uae_saudi_ai_governance.py |
UAE + Saudi Arabia (PDPL + AI Strategy 2031 + Saudi NDMO) | UAEPDPLFilter (Federal Decree-Law No. 45/2021: Art. 6 legal basis + Art. 9 sensitive health/financial/biometric/religious + Art. 22 cross-border DTA + Art. 13 automated decision), UAEAIRegFilter (UAE AI Strategy §4 impact assessment + DFSA RPP DIFC + FSRA AI RMF ADGM + UAE AI Ethics transparency), SaudiNDMOFilter (PDPL M/19: Art. 5 consent + Art. 23 sensitive health/genetic/biometric/financial/criminal + Art. 29 SDAIA authorization + NDMO DGF v2.0), GCCCrossBorderFilter (GCC adequacy + FATF KP/IR/MM financial + SDAIA export approval + GCC national cloud sovereignty), 8 ecosystem wrappers |
Agentic AI Security (OWASP 2026 + NIST + ISO + MITRE + CSA)
| 39 | 39_owasp_agentic_top10_governance.py | OWASP Agentic AI Top 10 2026 | OWASPAgenticInjectionFilter (ASI01 goal hijack + ASI04 prompt injection defense), OWASPAgenticToolSafetyFilter (ASI02 tool misuse + ASI05 scope creep + ASI06 resource abuse), OWASPAgenticDataLeakageFilter (ASI03 memory poisoning + ASI07 insecure output handling), OWASPAgenticGovernanceFilter (ASI08 trust boundaries + ASI09 audit gaps + ASI10 unvalidated composition), 8 ecosystem wrappers (LangChain/CrewAI/AutoGen/SK/LlamaIndex/Haystack/DSPy/MAF) |
| 40 | 40_governance_framework_auditor.py | Multi-Framework Governance Audit | Holistic governance gap-analysis — GovernanceFrameworkConfig (37 boolean controls), GovernanceFrameworkAuditor (37 controls spanning OWASP LLM Top 10 2025 (8), OWASP Agentic AI 2026 (4), NIST AI RMF + AI 600-1 (5), ISO/IEC 42001:2023 (9), MITRE ATLAS v5.1 (6), CSA ATF (5)), GovernanceAuditReport (scored 0–100, INITIAL/DEVELOPING/DEFINED/MANAGED/OPTIMIZING maturity, per-domain breakdown) |
| 41 | 41_trilogy_security_audit.py | Unified Enterprise AI Security Audit | TrilogyAuditOrchestrator — combines RAG + Agentic + Governance auditors into a single enterprise assessment: TrilogySystemProfile (35 fields across all 3 auditor dimensions), combined_score = RAG×0.35 + Agent×0.35 + Gov×0.30, combined_maturity = minimum of three (weakest link), 7 cross-module gap correlations (XG-001 injection mismatch → XG-007 HITL gap across layers), TrilogyAuditResult with box-formatted enterprise security report |
Jurisdiction and regulation coverage matrix
| Jurisdiction | Regulations Covered | Example |
|---|---|---|
| United States | FERPA, HIPAA, GLBA, NIST AI RMF, EO 14110, FINRA, SEC, NAIC | 01, 05, 07–09, 15 |
| European Union | EU AI Act (Art. 9/10/12/14), GDPR Art. 22 ADM, MiFID II, PSD2, DORA, Solvency II | 10, 13, 14, 16 |
| United Kingdom | UK GDPR Art. 22, ICO AI Auditing Framework 2022, Equality Act 2010 (s.19/s.149 PSED), DSIT AI Safety White Paper | 17 |
| Canada | AIDA Bill C-27, CPPA (replacing PIPEDA), OPC Guidelines 2023, Québec Law 25 (Act 25) | 18 |
| Australia | DIIS 8 AI Ethics Principles, Privacy Act 1988 (APPs 1/3/6/11, NDB), DTA ADM Framework, AHRC AI & Human Rights | 19 |
| Singapore | PDPC Model AI Governance Framework v2 (2020), PDPA 2012 (§§13/26/27/28), MAS FEAT (F.1/T.1/A.2), IMDA AI Testing | 20 |
| Japan | APPI 2022 (Art. 20-2/27/28), METI AI Governance Guidelines v1.1, MHLW Medical AI Guidelines, Cabinet Office Social Principles | 21 |
| Manufacturing / OT | IEC 62443, ISA-95, IEC 61508 SIL, ISO 26262 (automotive), UN-R155 | 06, 11 |
| MedTech / Healthcare | EU MDR, FDA 510(k), IEC 62304, HIPAA, FDA AI/ML Action Plan | 05, 12 |
| Brazil | LGPD Law 13.709/2018, Brazilian AI Bill PL 2338/2023, ANPD Guidelines 2023, CFM Resolution 2299/2021, BCB Circular 3.978/2020 | 22 |
| India | DPDP Act 2023, MEITY AI Advisory 2024, IT Act 2000 §43A + IT Rules 2011, RBI AI Guidance, IRDAI Guidelines, CDSCO SaMD Guidance | 23 |
| South Korea | Korea AI Framework Act 2024, PIPA 2020/2023 amendments, Credit Information Use and Protection Act, Korean Medical Devices Act (MFDS SaMD) | 24 |
| Middle East (GCC) | UAE PDPL Federal Decree-Law No. 45/2021 + UAE AI Ethics 2019, Saudi PDPL Royal Decree M/19 + NDMO + SAMA + SFDA, Qatar PDPA Law No. 13/2016 + Qatar AI Strategy 2030, GCC cross-border adequacy | 25 |
| Africa | Kenya DPA 2019 (§25/§30/§31), Nigeria NDPA 2023 (§25/§34) + NITDA AI Policy §3.2, South Africa POPIA 2013 (§26/§71) + FSCA AI Guidance 2023, AU Data Policy Framework 2022 cross-border adequacy | 26 |
| Nordic / Scandinavia | Sweden IMY AI Guidelines 2023 + SFS 2018:218, Denmark Datatilsynet AI Guidance 2023 + Act No. 502/2018, Finland TSV AI Guidelines 2023 + Data Protection Act 1050/2018, EEA intra-transfer adequacy | 28 |
| Eastern Europe (Poland, Czech Republic, Hungary) | Poland UODO AI Guidelines 2023 + GDPR Act Dz.U. 2018 poz. 1000, Czech ÚOOÚ AI Guidance 2023 + Act 110/2019 Coll., Hungary NAIH AI Guidelines 2023 + Privacy Act CXII/2011, 30-jurisdiction EEA adequate set + SCC/BCR fallback | 29 |
| US State AI Laws (Colorado, Illinois, Virginia) | Colorado AI Act 2024 SB 24-205 (impact assessment, human oversight, bias testing), Illinois BIPA 740 ILCS 14 (biometric consent, video interview AI), Virginia CDPA AI provisions §59.1-577/578/579 (sensitive data, automated profiling, high-risk AI), multi-state cross-border applicability matrix | 30 |
| EU AI Act — High-Risk Systems | EU AI Act 2024/1689 Art. 5 prohibited practices, Annex III §1-8 high-risk categories, Arts. 9-15 technical requirements, Arts. 50/13 transparency, Arts. 51-55 GPAI/systemic risk; 8 ecosystem wrappers (LangChain/CrewAI/AutoGen/SK/LlamaIndex/Haystack/DSPy/MAF) | 31 |
AI framework integrations
| Framework | Adapter | Notes |
|---|---|---|
| CrewAI | EnterpriseActionGuard |
Security quarantined; native support suspended |
| LangChain | GovernanceCallbackHandler |
callbacks.on_retriever_end |
| LlamaIndex | GovernedQueryEngine |
Policy checks around query execution |
| Google ADK | ADKPolicyGuard |
before_model_callback / before_tool_callback — BigQuery audit sink |
| AutoGen | Legacy object wrapper/API retained | Security quarantined; native SDK install unsupported |
| Semantic Kernel | GovernedKernelPlugin |
SK Plugin interface |
| Haystack 2.x | GovernedHaystackComponent |
Related standalone filter: ferpa-haystack on PyPI |
| DSPy | GovernedDSPyModule, GovernedDSPyPipeline |
Legacy wrapper; native SDK support not qualified |
| Microsoft Agent Framework (MAF) | PolicyMiddleware (as_native() bridge) |
Successor to AutoGen + Semantic Kernel |
Repository structure
src/regulated_ai_governance/
├── __init__.py # Core exports
├── policy.py # ActionPolicy, GovernanceAuditRecord
├── regulations/
│ ├── ferpa.py # FERPA 34 CFR §99
│ ├── hipaa.py # HIPAA 45 CFR §164
│ ├── gdpr.py # GDPR Art. 17/22
│ ├── glba.py # GLBA 16 CFR §314
│ └── ...
└── integrations/
├── crewai.py # EnterpriseActionGuard
├── langchain.py # GovernanceCallbackHandler
├── autogen.py # GovernedAutoGenAgent
├── llama_index.py # GovernedQueryEngine
├── semantic_kernel.py # GovernedKernelPlugin
├── haystack.py # GovernedHaystackComponent
├── dspy.py # GovernedDSPyModule, GovernedDSPyPipeline
└── maf.py # PolicyMiddleware
examples/ # Governance demonstrations (see catalog above)
tests/ # Regression tests; see CI for current results
ECOSYSTEM.md # Selected regulation/framework support boundaries
docs/ # Design, API and evidence documentation
Audit record format
Every filter evaluation produces a structured audit record:
{
"filter_name": "APPI_DATA_PROTECTION",
"decision": "DENIED",
"reason": "APPI Article 20-2: Opt-in consent required for sensitive personal information",
"regulation_citation": "Act on Protection of Personal Information (APPI) 2022 Article 20-2",
"requires_logging": True
}
JSON-serializable; suitable for compliance databases, SIEM ingestion, and regulator production requests.
Standalone Haystack Package
ferpa-haystack — FERPAMetadataFilter as a standalone Haystack 2.x custom component. Install with pip install 'ferpa-haystack>=0.4.0'. Adds identity-scope filtering to a configured Haystack pipeline. This separate package exports FERPAMetadataFilter, not this repository’s GovernedHaystackComponent. Supply identities and category grants from your authenticated authorization context; the values below are synthetic:
from haystack_integrations.components.filters.ferpa_filter import FERPAMetadataFilter
# Assumes a Haystack pipeline with a retriever already configured.
pipeline.add_component("ferpa_filter", FERPAMetadataFilter(
student_id="stu_001",
institution_id="inst_001",
authorized_categories=["academic_record"],
))
pipeline.connect("retriever.documents", "ferpa_filter.documents")
Google ADK Integration
ADKPolicyGuard — drop-in compliance layer for Google Agent Development Kit agents. Enforces FERPA, HIPAA, GDPR, GLBA, EU AI Act, and OWASP Agentic AI Top 10 2026 through ADK native before_model_callback, before_agent_callback, and before_tool_callback hooks. Supports BigQuery audit sink and multi-agent governance (Orchestrator → sub-agent authorization propagation).
from regulated_ai_governance.adapters.google_adk_adapter import ADKPolicyGuard, Regulation
guard = ADKPolicyGuard(regulations=[Regulation.FERPA, Regulation.HIPAA], agent_id="my-agent")
agent = LlmAgent(before_model_callback=guard.before_model_callback, ...)
Reliability and adoption
The source version and released distribution are separate artifacts. See CHANGELOG.md for unreleased changes; install a published version for reproducibility and record its version in deployment evidence.
- Audit guarantees and failure semantics
- Control provenance and applicability checklist
- Current roadmap
Contributing
Read CONTRIBUTING.md and GOVERNANCE.md. Run pytest tests/ -v before opening a pull request.
Citation
@software{rana2026rag,
author = {Rana, Ashutosh},
title = {regulated-ai-governance: Policy enforcement for AI agents in regulated environments},
year = {2026},
version = {0.47.1},
url = {https://github.com/ashutoshrana/regulated-ai-governance},
license = {MIT}
}
Part of the enterprise AI patterns trilogy
| Library | Focus | Coverage |
|---|---|---|
| enterprise-rag-patterns | What to retrieve | 50 sectors · 65 regulations · 1,901 tests |
| ferpa-haystack | Haystack-native FERPA filter | Standalone Haystack 2.x component · 25 tests |
| regulated-ai-governance | What agents may do | 41 governance examples · 25 jurisdictions · 2,631 tests |
| integration-automation-patterns | How data flows | 43 patterns · schema registry · GraphQL · 1,865 tests |
License
MIT — see LICENSE.
Metadata
Release files for regulated-ai-governance 0.47.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| regulated_ai_governance-0.47.1.tar.gz | 260.2 kB | With provenance | Details for regulated_ai_governance-0.47.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| regulated_ai_governance-0.47.1-py3-none-any.whl | Python 3 | none | any | Details for regulated_ai_governance-0.47.1-py3-none-any.whl |
Total release size: 378.5 kB
Release files / regulated_ai_governance-0.47.1.tar.gz
| Download URL | regulated_ai_governance-0.47.1.tar.gz |
|---|---|
| Size | 260.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
af0194d602d1e6d7bf8c772207ccc446116a9cda91a9facb5e17b071e77dd494
|
|
BLAKE2b-256 checksum How to use checksums |
0de431abcaf69816c70a7ca95bb0e2713b193456dcbf89b40138811bc1abbe9e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 13, 2026.
Transparency logRelease files / regulated_ai_governance-0.47.1-py3-none-any.whl
| Download URL | regulated_ai_governance-0.47.1-py3-none-any.whl |
|---|---|
| Size | 118.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
8d75d9c8579b4ffe48152f42c33d0f7452e658252f36a03df3c10bb9b49deb99
|
|
BLAKE2b-256 checksum How to use checksums |
19cb81a311ca1b7f74a102734366fec8db88f77972026cff1075db618c780590
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 13, 2026.
Transparency logSecurity for regulated-ai-governance 0.47.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for regulated_ai_governance-0.47.1.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 13, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.