Which countries require human oversight of AI?

Duties of human oversight or review, and rights to a human decision instead of an automated one.

20 of 109 reviewed jurisdictions address this in at least one tracked instrument. Updated 2026-10-04.

Each answer cites the instrument it rests on; open it for the provision and the full analysis. Whether a rule is binding law, a bill or guidance comes from the instrument's type and status, not from wording.

“Not found” means none of the instruments we track for that jurisdiction addresses the topic. It is not a finding that no such rule exists.

Yes — binding law in force (12)

Argentina
Yes — binding law in force
Generative-AI outputs require express validation and professional judgment by the responsible public official.
- Argentina - Santa Fe - AI Adoption Protocol (2726/2025), Anexo, Artículo 2° — Decree, In Force
- Argentina - San Luis - Judicial AI Protocol (2024), Sources and References / Acuerdo N° 202-STJSL-SA-2024 — Regulation, In Force
- Argentina - Virtual Assistant Implementation (1486/2025), Implementation requirements — Regulation, In Force
- and 7 more
Status verified 2026-09-09
4 of 4 regions address this
- Buenos AiresGuidance or policy only
  Administrative acts and decisions cannot rely solely on AI output without competent official review.
  - Argentina - Buenos Aires - Generative AI Guidelines, Definitions — Guideline, In Force
- JujuyGuidance or policy only
  Users must retain human responsibility and verify AI-generated outputs' legal and ethical compliance.
  - Argentina - Jujuy - AI Use Protocol (Acordada N° 31/2025), Implementation Framework — Guideline, In Force
  Status verified 2026-09-07
- Río NegroGuidance or policy only
  AI outputs must undergo human validation before use in judicial acts, and human officials retain decision responsibility.
  - Argentina - Río Negro - AI Good Practices Protocol (RAI-AR-RN-PDBPPXX-2024), Key Focus Areas — Guideline, In Force
  A possible change to a cited instrument is not yet verified.
- San JuanPartly — regional, framework or general (non-AI) law
  A human official must review, validate, and accept responsibility for AI output before legal effect.
  - Argentina - San Juan - AI Judicial Protocol (2024), Definitions — Regulation, In Force
  Status verified 2026-09-09
Brazil
Yes — binding law in force
Judicial computer systems must require human supervision and allow judges to modify AI-generated outputs.
- Brazil - AI Governance in Judiciary (615/2025), Art. 34 — Regulation, In Force
- Brazil - National AI Regulation (2.338/2023), Art. 3º, III — Bill, Under Review
- Brazil AI Regulatory Framework Bill, Art. 6º, III — Bill, Under Review
- and 1 more
A possible change to a cited instrument is not yet verified.
China
Yes — binding law in force
A human must take over conversations when users explicitly present suicide or self-harm scenarios.
- China - Anthropomorphic AI Services Regulation, 第十一条 — Regulation, In Force
- China - Algorithmic Recommendation Regulation (Order No. 9), 第十一条 — Regulation, In Force
- China - Personal Information Protection Law (2021) — Act, In Force
- and 2 more
Status verified 2026-09-10
0 of 4 regions address this
- BeijingNot found in the instruments we track
- ShanghaiNot found in the instruments we track
- ShenzhenNot found in the instruments we track
- ZhejiangNot found in the instruments we track
Finland
Yes — binding law in force
High-risk AI systems must allow effective intervention and oversight by natural persons.
- Finland AI Act Implementation Law, Compliance Checklist — Act, In Force
- Finland - AI Impact on Work (21/2018), Key Focus Areas — Policy, Adopted
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
1 of 1 regions address this
- HelsinkiGuidance or policy only
  Registered systems must have structured human oversight arrangements for reviewing and overriding outputs.
  - Finland - Helsinki - AI Register, Definitions — Policy, In Force
  Status verified 2026-09-06
Hungary
Yes — binding law in force
Providers and deployers must comply with EU AI Regulation obligations including human oversight.
- Hungary - AI Regulation Implementation (344/2025) — Decree, In Force
- Hungary - EU AI Regulation Implementation (LXXV/2025) — Act, In Force
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
Italy
Yes — binding law in force
General-purpose AI systems and models must ensure human supervision and intervention.
- Italy - AI Regulation (n.132/2025), Art. 3, comma 3 — Act, In Force
- Italy - AI Adoption Guidelines, §3.4 — Guideline, Draft
- Italy - Digital Transformation Plan, § 5.5.5 Come affrontare l’AI Act — Policy, In Force
Status verified 2026-09-08
As an EU member state, the EU AI Act also applies here.
1 of 2 regions address this
- Emilia-RomagnaGuidance or policy only
  AI solutions must comply with the Italian AI Law and EU AI Act, with a stated focus on human oversight.
  - Italy - Emilia-Romagna - AI Regional Hub — Policy, In Force
  Status verified 2026-09-09
- LombardyNot found in the instruments we track
Kazakhstan
Yes — binding law in force
Owners, possessors and users must maintain continuous control over AI throughout its lifecycle.
- Kazakhstan - AI Regulation (2025), Статья 8, пункт 1 — Act, In Force (Amended)
- Kazakhstan Digital Code, Penalties, Liability, and Appeals — Act, Adopted (from 2026-07-01)
- Kazakhstan AI Ethics Code, Definitions — Guideline, In Force
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-10
Malta
Yes — binding law in force
Operators must ensure human oversight measures and user information notices are in place.
- Malta - AI Regulations (226/2025), Key Focus Areas — Regulation, In Force
- Malta - AI Commissioner Designation (227/2025), Implementation Framework — Regulation, In Force
- Malta - Ethical AI Framework, Compliance Checklist — Guideline, Adopted (from 2019-10-03)
Status verified 2026-09-08
As an EU member state, the EU AI Act also applies here.
Peru
Yes — binding law in force
High-risk AI systems must have human supervision for decisions with significant impacts in specified sectors.
- Peru - AI Regulation Implementation (115-2025), Artículo 28.11; Artículo 31.4 — Decree, In Force
- Peru - AI Interoperability for Foreign Affairs (07619), Key Focus Areas — Act, In Force
- Peru - AI Education Integration (7659/2023), Key Focus Areas — Bill, Under Review
Status verified 2026-09-09
Singapore
Yes — binding law in force
Court users must personally fact-check, proof-read and adapt AI-generated material before filing.
- Singapore - Generative AI Use in Courts (Registrar's Circular No. 1/2024), Core obligations — Regulation, In Force
- Singapore - Ethical AI Advisory Council — Guideline, In Force
- Singapore - AI Model Risk Management, Section 7.1.9 — Guideline, In Force
- and 2 more
Status verified 2026-09-10
Slovakia
Yes — binding law in force
High-risk AI systems must include human oversight, and deployers must ensure and monitor that oversight.
- Slovakia AI Conformity Assessment Act, Key Focus Areas — Act, In Force
- Slovakia - AI Governance Bill (LP/2025/401), Key Focus Areas — Bill, Under Review
- Slovakia - AI State Administration Act (RAI-SK-NA-DOSAFXX-2025) — Bill, Under Review
- and 2 more
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
Vietnam
Yes — binding law in force
High-risk AI systems must have effective human oversight and control.
- Vietnam AI Act 2025 — Act, In Force
- Vietnam - Responsible AI Development (1290/QD-BKHCN), Key Focus Areas — Controllability — Guideline, In Force
Status verified 2026-09-08
0 of 1 regions address this
- Ho Chi Minh CityNot found in the instruments we track

Partly — regional, framework or general (non-AI) law (6)

Estonia
Partly — regional, framework or general (non-AI) law
Automated-decision systems must provide for human intervention when errors or exceptions occur.
- Estonia - Automated Administrative Acts (2018), §33⁶ — Act, In Force
- Estonia - Data and AI Strategy, Implementation expectations — Policy, Draft
- Estonia - National AI Strategy (2019–2021), Compliance Checklist — Policy, Adopted (from 2019-07-25)
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
India
Partly — regional, framework or general (non-AI) law
Intermediaries must use proportionate technology-based content-identification measures with human oversight and review of automated tools.
- India - Digital Media Ethics Code (2021), Implementation Framework — Regulation, In Force (Amended)
- India - AI Governance Guidelines (2025), Part 2, 2.4 Risk Mitigation, Recommendations — Guideline, In Force
- India - National AI Strategy, Key Focus Areas — Policy, Adopted (from 2018-06-01)
Status verified 2026-09-08
1 of 4 regions address this
- KarnatakaNot found in the instruments we track
- MaharashtraGuidance or policy only
  AI advisories are decision-support tools and do not replace expert judgment or farmer responsibility.
  - India - Maharashtra - AI in Agriculture (2025-29), Penalties, Liability, and Appeals — Policy, In Force
  Status verified 2026-09-09
- Tamil NaduNot found in the instruments we track
- TelanganaNot found in the instruments we track
Nigeria
Partly — regional, framework or general (non-AI) law
Large Service Platforms must provide human oversight of automated moderation systems.
- Nigeria - Code of Practice for Platforms (2022) — Regulation, In Force
- Nigeria - AI Technology Control (HB 942) — Bill, Under Review
- Nigeria - AI Code of Practice (2025), Compliance obligations — Regulation, Draft
- and 2 more
Status verified 2026-09-07
Spain
Partly — regional, framework or general (non-AI) law
AI systems must be designed for supervision by natural persons while in use.
- Spain - National AI Sandbox (Royal Decree 817/2023), Artículo 11.g — Decree, In Force
- Spain AI Governance Bill, Sistemas de alto riesgo — Bill, Under Review
- Spain - Digital Rights Framework — Policy, Adopted
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
3 of 3 regions address this
- AsturiasPartly — regional, framework or general (non-AI) law
  Automated administrative decisions must be reviewable, and affected persons have a right to human intervention.
  - Spain - Asturias - AI Public Administration Decree (2025), artículo 3.h — Decree, In Force
  - Spain - Asturias - AI Regulation (98/2025) — Decree, In Force
  Status verified 2026-09-09
- CataloniaGuidance or policy only
  The compliance checklist calls for individuals to retain meaningful control over AI systems.
  - Spain - Catalonia - Ethical AI Framework, Compliance Checklist — Policy, In Force
  Status verified 2026-09-09
- GaliciaPartly — regional, framework or general (non-AI) law
  AI-driven administrative decisions must allow human supervision, intervention, and review.
  - Spain - Galicia - AI Regulation (2/2025), Compliance Checklist — Act, In Force
  - Spain - Galicia - AI Development Law (2/2025), Artículo 23.1.c — Act, In Force (Amended)
  A possible change to a cited instrument is not yet verified.
United Kingdom
Partly — regional, framework or general (non-AI) law
Automated decisions using non-sensitive personal data must include human review pathways.
- United Kingdom - Data Access Reform (2025) — Act, In Force
- United Kingdom - AI Playbook, Human oversight — Guideline, In Force (Amended)
- United Kingdom - Defence AI Strategy, Compliance Checklist — Policy, In Force
- and 1 more
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-09
0 of 1 regions address this
- ScotlandNot found in the instruments we track
United States
Partly — regional, framework or general (non-AI) law
Coverage determinations using AI must be individualized rather than based solely on algorithmic outputs.
- CMS 2024 Medicare Advantage Policy and AI Guidance, Overview — Regulation, In Force
- US Federal AI Acceleration Memo, Section 4(b)(v) — Policy, In Force
- ABA Guidance on Lawyers' Ethical AI Use, Key Focus Areas — Competence (Model Rule 1.1) — Guideline, In Force
- and 6 more
Status verified 2026-09-09
14 of 51 states address this
- AlabamaNot found in the instruments we track
- AlaskaNot found in the instruments we track
- ArizonaPartly — regional, framework or general (non-AI) law
  Insurers must obtain an individual medical-director review exercising independent judgment before specified medical-necessity denials.
  - United States - Arizona - AI Claim Denial Restrictions (HB 2175), A.R.S. §§ 20-3103 and 20-3407 — Act, In Force
  Status verified 2026-09-09
- ArkansasPartly — regional, framework or general (non-AI) law
  Public-entity policies must preserve final decision authority for human employees or designees.
  - United States - Arkansas - AI Policy Requirements (Act 848), Key Focus Areas — Act, In Force
  Status verified 2026-09-07
- CaliforniaPartly — regional, framework or general (non-AI) law
  Driverless testing requires continuous monitoring and a remote operator able to take control in emergencies.
  - California AV Testing and Deployment Rules, Key Focus Areas — Regulation, In Force
  - California Healthcare AI Transparency and Accountability Bill, Labor Code Section 2821(b)(1) — Bill, Adopted (from 2027-07-01)
  Status verified 2026-09-08
- ColoradoPartly — regional, framework or general (non-AI) law
  A regulated professional must retain oversight and responsibility for administrative AI use and approve treatment recommendations.
  - Colorado Psychotherapy AI Restrictions Act — Act, In Force
  - Colorado AI in Healthcare Act, Implementation Framework — Act, Awaiting Entry (from 2027-01-01)
  - Colorado Automated Decision-Making Technology Act — Act, Awaiting Entry (from 2027-01-01)
  Status verified 2026-09-08
- ConnecticutNot found in the instruments we track
- DelawareNot found in the instruments we track
- FloridaNot found in the instruments we track
- GeorgiaAdopted — not yet in force
  Final healthcare coverage denials must be reviewed and approved by a qualified human clinical peer.
  - Georgia AI in Insurance Coverage Act — Act, Awaiting Entry (from 2027-01-01)
  - United States - Georgia - AI Governance Guidelines (2024), Implementation Framework — Guideline, In Force (Amended)
  - United States - Georgia - AI in Courts — Policy, Adopted (from 2025-07-03)
  Status verified 2026-09-07
- HawaiiNot found in the instruments we track
- IdahoPartly — regional, framework or general (non-AI) law
  AI implementations must preserve human educators' ultimate control and decision-making authority.
  - Idaho K-12 Generative AI Education Act, Key Focus Areas — Act, In Force
  Status verified 2026-09-10
- IllinoisPartly — regional, framework or general (non-AI) law
  Licensed professionals must review and approve AI therapeutic recommendations before delivery to clients.
  - Illinois AI in Therapy Act — Act, In Force
  - Illinois AI Teacher Evaluation Ban, Overview — Bill, Adopted
  - Illinois AI Healthcare Downcoding Bill, Implementation Framework — Bill, Awaiting Entry (from 2028-01-01)
  - and 1 more
  A possible change to a cited instrument is not yet verified.
  Status verified 2026-09-08
- IndianaPartly — regional, framework or general (non-AI) law
  Insurers and providers must ensure human review when automated processes, including AI, are used for specified claims actions.
  - Indiana AI Health Claims Law, IC 27-1-52 — Act, In Force
  A possible change to a cited instrument is not yet verified.
  Status verified 2026-09-09
- IowaNot found in the instruments we track
- KansasNot found in the instruments we track
- KentuckyNot found in the instruments we track
- LouisianaNot found in the instruments we track
- MaineNot found in the instruments we track
- MarylandNot found in the instruments we track
- MassachusettsNot found in the instruments we track
- MichiganNot found in the instruments we track
- MinnesotaNot found in the instruments we track
- MississippiNot found in the instruments we track
- MissouriNot found in the instruments we track
- MontanaPartly — regional, framework or general (non-AI) law
  AI-controlled critical infrastructure must allow human control and intervention during emergencies.
  - United States - Montana - Right to Compute Act (SB 212) — Act, In Force
  Status verified 2026-09-08
- NebraskaPartly — regional, framework or general (non-AI) law
  Adverse prior-authorization determinations must be made by a qualified physician or clinical peer rather than automated systems.
  - Nebraska Prior Authorization Transparency Act — Act, In Force
  Status verified 2026-09-08
- NevadaNot found in the instruments we track
- New HampshireNot found in the instruments we track
- New JerseyNot found in the instruments we track
- New MexicoNot found in the instruments we track
- New YorkNot found in the instruments we track
- North CarolinaGuidance or policy only
  Human oversight is mandatory throughout AI development, deployment, and use.
  - United States - North Carolina - Responsible AI Framework, Key Focus Areas — Policy, In Force
  Status verified 2026-09-17
- North DakotaNot found in the instruments we track
- OhioNot found in the instruments we track
- OklahomaNot found in the instruments we track
- OregonNot found in the instruments we track
- PennsylvaniaNot found in the instruments we track
- Puerto RicoNot found in the instruments we track
- Rhode IslandNot found in the instruments we track
- South CarolinaNot found in the instruments we track
- South DakotaNot found in the instruments we track
- TennesseeNot found in the instruments we track
- TexasNot found in the instruments we track
- UtahProposed
  Insurers must ensure adverse clinical or medical-necessity determinations use independent medical judgment rather than solely another source's recommendations.
  - Utah Health Insurance AI Disclosure Law, Section 31A-22-650(6)(e)(ii) — Bill, In Force
  Status verified 2026-09-07
- VermontGuidance or policy only
  AIS programmes should assign senior management responsibility for monitoring and oversight of the programme.
  - United States - Vermont - AI in Insurance (DFR Bulletin 229), Section 3, AIS Program Guidelines, 1.3 — Guideline, In Force
  Status verified 2026-09-06
- VirginiaNot found in the instruments we track
- WashingtonPartly — regional, framework or general (non-AI) law
  AI cannot be the sole basis for adverse care decisions, and medical-necessity denials require human review.
  - Washington AI in Insurance Prior Authorization Act, Sec. 2, RCW 48.43.830(3)(a) — Act, In Force
  - United States - Washington - Responsible AI Policy, Implementation Framework — Policy, In Force
  Status verified 2026-09-07
- West VirginiaNot found in the instruments we track
- WisconsinNot found in the instruments we track
- WyomingNot found in the instruments we track

Proposed (21)

Bahrain
Proposed
Human intervention or assessment is mandatory for AI decisions significantly affecting people, life, freedom, or bodily integrity.
- Bahrain - AI Regulation Law (2024) — Bill, Under Review
Status verified 2026-09-08
Bangladesh
Proposed
Social media intermediaries must provide human oversight and periodic review for automated content-identification tools.
- Bangladesh - Digital Media Regulation (draft), Clause 7.04 — Regulation, Draft
- Bangladesh - National AI Policy (2024), Governance — Policy, Draft
Status verified 2026-09-09
Chile
Proposed
The bill mandates human oversight and identification of human-in-the-loop points.
- Chile - AI and Robotics Regulation (Boletín 15869-19) — Bill, Under Review
- Chile - AI Development Limits (17112-19), Antecedentes — Bill, Under Review
- Chile - AI Systems Regulation (Boletín 16821-19) — Bill, Under Review
- and 1 more
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-08
Colombia
Proposed
The proposal requires controls enabling contestability and human review of automated decisions.
- Colombia - Ethical AI Law Proposal (PL 005/2024C) — Bill, Stalled
- Colombia - AI Regulation Bill (C200/2023) — Bill, Stalled
- Colombia - AI Regulation Framework (Senate 442/25), Compliance Checklist — Bill, Stalled
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-10
Costa Rica
Proposed
Professional and technical judgment must take priority over automated AI decision systems.
- Costa Rica AI Regulation Bill, ARTÍCULO 8 — Bill, Under Review
Status verified 2026-09-08
Croatia
Proposed
Operators must design and document human-in-the-loop controls.
- Croatia - AI Act Transposition Process (2025) — Bill, Draft
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
Czech Republic
Proposed
The text directs operators to plan human-oversight measures.
- Czech Republic - AI Implementation Act — Bill, Draft
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
Ecuador
Proposed
High-stakes AI decisions require meaningful human involvement, and affected persons have human-review rights.
- Ecuador - AI Regulation and Promotion (2024) — Bill, Under Review
Status verified 2026-09-09
France
Proposed
Algorithmic treatments must remain under human control and their alerts must be assessed by trained, authorized officers.
- France - Biometric Recognition Bill (n°505), Article 5, IV — Bill, Under Review
Status verified 2026-09-08
As an EU member state, the EU AI Act also applies here.
Indonesia
Proposed
Human oversight is required for AI systems making critical decisions.
- Indonesia - National AI Regulation, Obligations and enforcement — Decree, Draft
- Indonesia - National AI Roadmap, Compliance Checklist — Policy, Draft
- Indonesia - AI Governance for Banking, Key Focus Areas — Human oversight — Guideline, In Force
- and 2 more
Status verified 2026-09-08
Luxembourg
Proposed
It requires human oversight for high-risk AI systems.
- Luxembourg - AI Act Implementation (Bill No. 8476) — Act, Under Review
- Luxembourg - AI in Schools Guidelines, Implementation requirements — Policy, Draft
- Luxembourg - National AI Strategy (2025), Implementation Framework — Policy, Adopted
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
Morocco
Proposed
AI-assisted automated decisions affecting individuals must provide human review rights.
- Morocco - Digital Services Framework (2025), Key Focus Areas — Bill, Draft
A possible change to a cited instrument is not yet verified.
1 of 1 regions address this
- Casablanca-SettatGuidance or policy only
  AI solution design, development and deployment must integrate human oversight as an ethical consideration.
  - Morocco - Casablanca-Settat - AI Excellence Center — Policy, In Force
  Status verified 2026-09-08
Netherlands
Proposed
High-risk AI must allow natural persons to oversee and intervene in its operation.
- Netherlands AI Act Implementation Draft, Key Focus Areas — Bill, Proposed
- Netherlands - AI Oversight Guidelines, 5.7 Specifieke normen voor uitsluitend geautomatiseerde besl — Guideline, In Force
- Netherlands - AI and Human Rights (2019), staged approach / government use of AI — Policy, Adopted
- and 8 more
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-08
As an EU member state, the EU AI Act also applies here.
1 of 1 regions address this
- AmsterdamGuidance or policy only
  Systems significantly affecting individuals require human review before a final decision.
  - Netherlands - Amsterdam - Algorithm Transparency (2020), Definitions — Policy, In Force
  A possible change to a cited instrument is not yet verified.
  Status verified 2026-09-09
Norway
Proposed
Operators of high-risk systems must define human-in-the-loop controls and receive training.
- Norway - AI Regulation (2026), Compliance Checklist — Bill, Draft
- Norway - National Digitalisation Strategy — Policy, Adopted (from 2027-01-01)
- Norway - National AI Strategy — Policy, Adopted (from 2020-01-14)
Status verified 2026-09-10
Panama
Proposed
It requires human involvement for high-stakes AI decisions affecting citizens.
- Panama - AI Regulation (Bill No. 14), Key Focus Areas — Bill, Under Review
A possible change to a cited instrument is not yet verified.
Philippines
Proposed
It requires meaningful human control over AI systems used in critical applications.
- Philippines - AI Development Regulation (HB 7396), Key Focus Areas — Ethical AI Principles and Implementation — Bill, Under Review
- Philippines - AI Labor Protection (HB 9448), House Bill 9448 includes important exceptions — Bill, Under Review
Status verified 2026-09-08
Poland
Proposed
High-risk AI systems must have appropriate human oversight.
- Poland - AI Systems Regulation (DAISPXX/2024) — Bill, Draft
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
Thailand
Proposed
Organizations must implement human oversight when it is necessary as part of risk management.
- Thailand - AI Risk Assessment Guidelines, Key Focus Areas — Regulation, Draft
- Thailand - AI Business Operations (2022), Key Focus Areas — Decree, Draft
- Thailand - AI Innovation Testing Center, Key Focus Areas — Regulation, Draft
- and 4 more
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-08
Turkey
Proposed
High-risk AI uses must have mandatory human oversight.
- Turkey - AI Law Amendments (Esas No. 2/3358), Key Focus Areas — Bill, Under Review
- Turkey - GAI Ethics Guide, Governance and Institutional Framework — Guideline, In Force
- Turkey - Personal Data Protection Guidelines, Key Focus Areas — Guideline, In Force
- and 1 more
Status verified 2026-09-09
Ukraine
Proposed
It requires meaningful human authority, intervention capability, and retained human decision-making over consequential AI.
- Ukraine - AI and Human Rights, Key Focus Areas — Human Oversight and Control — Regulation, Proposed
- Ukraine - AI Development Strategy (Order No. 1556-р) — Policy, In Force (Amended)
- Ukraine - Responsible AI in Advertising, TEXT — requirements list — Guideline, In Force
- and 1 more
Status verified 2026-09-10
Uzbekistan
Proposed
Decision-makers may not rely solely on AI conclusions for decisions affecting citizens' rights and freedoms.
- Uzbekistan - AI Regulation (2025) — Bill, Proposed
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-10

Guidance or policy only (32)

Australia
Guidance or policy only
The implementation measures require human oversight proportionate to risk.
- Australia - AI Policy Update, Mandatory requirements and implementation measures — Policy, In Force
- Australia - AI Ethics Framework (2019), Executive summary — Guideline, Adopted (from 2019-11-07)
- Australia - AI Assurance Framework (2024), Compliance Checklist — Guideline, In Force
- and 1 more
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-09
7 of 7 regions address this
- Australian Capital TerritoryGuidance or policy only
  AI-assisted decisions must have human oversight enabling review and intervention.
  - Australia - Australian Capital Territory - AI Policy (2025), Implementation requirements — Policy, In Force
  Status verified 2026-09-09
- New South WalesGuidance or policy only
  High- or critical-risk projects must receive review by the NSW AI Review Committee.
  - Australia - New South Wales - AI Assurance Framework (2022), Implementation Framework — Policy, In Force
  Status verified 2026-09-17
- Northern TerritoryGuidance or policy only
  All AI systems and projects must undergo assessment throughout their lifecycle, with regular reviews of AI services.
  - Australia - Northern Territory - AI Assurance Framework (2024), Implementation Framework — Policy, In Force
  Status verified 2026-09-09
- QueenslandGuidance or policy only
  Agencies must apply an ethical framework that addresses human oversight.
  - Australia - Queensland - AI Governance Policy (2024), Key Focus Areas — Policy, In Force
  Status verified 2026-09-10
- TasmaniaGuidance or policy only
  Agencies are advised to retain human oversight and accountability for AI outcomes.
  - Australia - Tasmania - AI Use Guidelines (2024), Penalties, Liability, and Appeals — Guideline, In Force
  Status verified 2026-09-08
- VictoriaGuidance or policy only
  Human officers must oversee checks of AI-created records and approve disposal of AI-derived records.
  - Australia - Victoria - AI Recordkeeping Policy (RAI-AU-VI-AIATRXX-2024), Key Focus Areas — Policy, In Force
  - Australia - Victoria - Generative AI Guidelines, Implementation support and technical guidance — Guideline, In Force
  Status verified 2026-09-08
- Western AustraliaGuidance or policy only
  Each health entity must appoint an executive officer to oversee AI governance.
  - Australia - Western Australia - AI Policy Framework (RAI-AU-WA-WESAUAI-2024), Section 3.3 Governance — Policy, In Force
  Status verified 2026-09-09
Belgium
Guidance or policy only
The roadmap promotes human oversight in AI procurement and deployment.
- Belgium - National AI Coalition, Key Focus Areas — Policy, Adopted (from 2019-03-01)
- Belgium - AI for Common Good — Policy, In Force
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
Bulgaria
Guidance or policy only
AI outputs affecting learning or assessment must remain subject to qualified educator intervention and override.
- Bulgaria - AI in Education Strategy, Definitions — Policy, Draft
Status verified 2026-09-08
As an EU member state, the EU AI Act also applies here.
Canada
Guidance or policy only
Institutions must retain human accountability for AI outputs.
- Canada - Generative AI Guidance, FASTER principles — Guideline, In Force
- Canada - Automated Decision-Making Directive, 6.3.9 — Policy, In Force (Amended)
- Canada - Accessible AI Standard (CAN-ASC-6.2:2025), Key requirements across the AI lifecycle — Standard, In Force
- and 3 more
Status verified 2026-09-08
3 of 4 regions address this
- AlbertaNot found in the instruments we track
- British ColumbiaGuidance or policy only
  Human roles and responsibilities must be defined for AI design, deployment, and oversight.
  - Canada - British Columbia - AI Responsible Use Principles, Compliance Checklist — Policy, Draft
  Status verified 2026-09-10
- OntarioPartly — regional, framework or general (non-AI) law
  Public sector entities must ensure an individual oversees the use of an AI system.
  - Canada - Ontario - AI Governance Act (Bill 194), Key Focus Areas — AI System Requirements — Act, In Force
  Status verified 2026-09-09
- QuebecGuidance or policy only
  AI responsibility entails controls and appropriate governance, including human supervision or validation.
  - Canada - Quebec - Responsible AI Use Guidelines, Principe de la responsabilité 9 — Guideline, In Force
  A possible change to a cited instrument is not yet verified.
Cyprus
Guidance or policy only
It recommends that public-sector AI pilots include redress and human-oversight mechanisms.
- Cyprus - National AI Strategy, Penalties, Liability, and Appeals — Policy, Adopted
A possible change to a cited instrument is not yet verified.
As an EU member state, the EU AI Act also applies here.
Denmark
Guidance or policy only
The cited non-binding AI ethics guidelines state that AI systems should be controllable and overseen by humans.
- Denmark - AI Regulation Supplement (Law No. 467/2025), Recital (27) — Act, In Force
- Denmark - AI Use by Public Authorities, 3.6.2 Hvad er automatiske afgørelser? — Guideline, In Force
- Denmark - Responsible Generative AI Guidelines, Key Focus Areas — Guideline, In Force
- and 1 more
Status verified 2026-09-08
As an EU member state, the EU AI Act also applies here.
Egypt
Guidance or policy only
Humans must retain ultimate authority and be able to modify, stop, or retire AI systems.
- Egypt - Responsible AI Guidelines, Key Focus Areas — Human Oversight — Guideline, Adopted (from 2023-04-19)
Status verified 2026-09-10
Ethiopia
Guidance or policy only
Human oversight is included in non-binding guidelines and as a safeguard for risk-based governance.
- Ethiopia - National AI Policy (2024), Key Focus Areas (4) Trustworthy AI and ethics — Policy, In Force
Status verified 2026-09-09
Germany
Guidance or policy only
The Roadmap recommends accelerating standards development on human oversight.
- Germany - AI Standardisation Roadmap, Implementation Framework — Policy, Adopted (from 2023-01-26)
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
2 of 3 regions address this
- Baden-WürttembergGuidance or policy only
  Projects seeking Research Fund support must undergo ethical review by the Public Advisory Board.
  - Germany - Baden-Württemberg - AI Research Initiative — Policy, In Force
  Status verified 2026-09-09
- BavariaNot found in the instruments we track
- North Rhine-WestphaliaGuidance or policy only
  The compliance checklist directs that AI systems allow human agency and oversight.
  - Germany - North Rhine-Westphalia - AI Strategy 2.0, Compliance Checklist — Policy, In Force
Greece
Guidance or policy only
It recommends fallback mechanisms allowing automated decisions to be overridden by humans.
- Greece - National AI Strategy, Appendix I — Policy, Under Review
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
Ireland
Guidance or policy only
All public-service AI tools must operate within a process containing human oversight.
- Ireland - Responsible AI Use Guidelines, Chapter 4, 4.1.2, Human oversight — Guideline, In Force
As an EU member state, the EU AI Act also applies here.
Israel
Guidance or policy only
It recommends human-centred design and human-in-the-loop measures for systems affecting rights.
- Israel - AI Regulation and Ethics, Key Focus Areas — Policy, Adopted (from 2023-12-14)
- Israel AI Regulation in Financial Sector Report, Compliance Checklist — Policy, Draft
- Israel - Algorithmic Processing Guidance, Key Focus Areas — Guideline, In Force
- and 1 more
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-09
Japan
Guidance or policy only
Human-in-the-loop controls should be used where AI poses significant risks to people.
- Japan - AI Utilization Guidelines, Key Focus Areas (6) — Guideline, Adopted (from 2019-08-09)
- Japan - AI R&D Guidelines, Key Focus Areas — Guideline, Draft
- Japan's National AI Basic Plan, Compliance Checklist — Policy, In Force (Amended)
Status verified 2026-09-10
Kenya
Guidance or policy only
The Code recommends human oversight and control measures suited to system risk and context.
- Kenya - AI Code of Practice (DKS 3007:2024), Compliance Checklist — Standard, Adopted (from 2024-09-30)
- Kenya - Media Reporting Guidelines, Key Focus Areas — Guideline, Adopted
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-09
Lebanon
Guidance or policy only
The checklist requires ethical guidelines for AI development and deployment that focus on human oversight.
- Lebanon - AI Initiative, Compliance Checklist — Policy, In Force
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-09
Lithuania
Guidance or policy only
It recommends that high-risk AI systems enable natural persons to oversee their functioning.
- Lithuania - AI Development Amendments (XV-105), (73) — Act, In Force
- Lithuania - AI Use in Public Sector, Key Focus Areas — Policy, Adopted (from 2024-05-09)
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
Malaysia
Guidance or policy only
Affected individuals must be provided meaningful human-oversight options for automated decisions.
- Malaysia National AI Action Plan 2030 — Policy, In Force
- Malaysia - National AI Roadmap, Compliance Checklist — Policy, In Force
- Malaysia - AI Governance Guidelines, AIGE risk-based approach — Guideline, In Force
Status verified 2026-09-08
New Zealand
Guidance or policy only
It recommends human oversight and review for AI decisions affecting people.
- New Zealand - AI Privacy Guidance, Overview — Guideline, In Force
- New Zealand - Algorithm Use Guidelines, Key Focus Areas — Guideline, In Force
- New Zealand - AI Strategy (2025) — Policy, Adopted (from 2025-07-08)
- and 4 more
Status verified 2026-09-06
Pakistan
Guidance or policy only
Public-sector AI procurement must include human-oversight clauses.
- Pakistan - National AI Policy (2025), Implementation Framework — Policy, Adopted (from 2025-07-30)
Status verified 2026-09-09
Portugal
Guidance or policy only
It recommends human-in-the-loop or human-on-the-loop oversight according to risk.
- Portugal - Ethical AI in Public Administration, Key Focus Areas — Guideline, In Force
Status verified 2026-09-09
As an EU member state, the EU AI Act also applies here.
Qatar
Guidance or policy only
It recommends human intervention mechanisms for high-impact decisions.
- Qatar - AI Security Guidelines, Practical recommendations — Guideline, In Force
- Qatar - Ethical AI Guidelines, Key Focus Areas — Guideline, In Force
- Qatar - AI Guidelines for Banks, Overview; Key Focus Areas — Guideline, In Force
1 of 1 regions address this
- DohaGuidance or policy only
  A human oversight protocol is mandatory for every AI system.
  - Qatar - AI Guidelines for Finance (2024), Implementation Framework — Guideline, In Force
  Status verified 2026-09-09
Romania
Guidance or policy only
Its human-centred AI principle places AI under human intelligence and action, with humans as final decision-makers.
- Romania - National AI Strategy, CAPITOLUL 2, secțiunea 2.3 — Policy, Adopted
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
Russia
Guidance or policy only
It recommends comprehensive human supervision and possible human cancellation or prevention of significant AI actions.
- Russia - AI Ethics Guidelines (2021), Section I, 3.1 — Guideline, Adopted (from 2021-10-26)
Status verified 2026-09-09
0 of 2 regions address this
- MoscowNot found in the instruments we track
- TatarstanNot found in the instruments we track
Rwanda
Guidance or policy only
The compliance checklist requires meaningful human control over AI-assisted decisions.
- Rwanda - National AI Policy, Compliance Checklist — Policy, Adopted (from 2023-04-20)
Status verified 2026-09-10
Saudi Arabia
Guidance or policy only
Human oversight is required for GenAI outputs used in critical public-service domains.
- Saudi Arabia - Generative AI Guidelines, Key Focus Areas — Guideline, In Force
- Saudi Arabia - AI Ethics Principles — Guideline, In Force
Status verified 2026-09-09
Serbia
Guidance or policy only
AI systems should remain under constant human monitoring and control.
- Serbia - Ethical AI Guidelines, 4.1 Human agency and control — Guideline, Adopted (from 2023-03-23)
Status verified 2026-09-09
South Africa
Guidance or policy only
It supports a human-in-command approach in which people make ultimate work-related conclusions.
- South Africa - Fourth Industrial Revolution Report — Policy, Adopted
South Korea
Guidance or policy only
Controllers must enable refusal of solely automated harmful decisions and provide human review.
- South Korea - Automated Decision Rights Guide — Guideline, Adopted
- South Korea - User Protection for Generative AI Services, Ⅳ. 기본 원칙 1 — Guideline, In Force
- South Korea - Public Data Processing Guide, Implementation measures — Guideline, In Force
Status verified 2026-09-08
0 of 1 regions address this
- SeoulNot found in the instruments we track
Sweden
Guidance or policy only
It recommends human-in-the-loop principles in public-sector AI procurement and operations.
- Sweden - AI Ecosystem Roadmap (SOU 2025:12), Recommendations — Policy, Under Review
Status verified 2026-09-10
As an EU member state, the EU AI Act also applies here.
Switzerland
Guidance or policy only
It recommends that AI deployments in government and critical sectors include human oversight where rights are affected.
- Switzerland - AI Challenges Report — Policy, Adopted
- Switzerland - AI Governance Guidelines (08/2024), Key Focus Areas — Guideline, In Force
- Switzerland - Generative AI Guidelines, Core rules — Guideline, In Force
- and 1 more
Status verified 2026-09-08
Taiwan
Guidance or policy only
The human-autonomy principle provides that ultimate decision-making power should remain with humans.
- Taiwan AI Basic Act — Act, In Force
- Taiwan - AI Evaluation Centre, Key Focus Areas — Policy, In Force
Status verified 2026-09-09
0 of 1 regions address this
- KaohsiungNot found in the instruments we track
United Arab Emirates
Guidance or policy only
It requires human-in-the-loop or human-on-the-loop controls for AI systems.
- United Arab Emirates - AI Adoption Guidelines, Key Focus Areas — Guideline, In Force
- United Arab Emirates - AI Ethics Guidelines, Key Focus Areas — Guideline, In Force
- United Arab Emirates - AI Ethics Self-Assessment Tool, Key Focus Areas — Guideline, In Force
- and 2 more
1 of 2 regions address this
- Abu DhabiNot found in the instruments we track
- DubaiGuidance or policy only
  Critical AI decisions require review and approval by a human operator.
  - United Arab Emirates - Dubai - AI Certification Framework, Definitions — Policy, In Force
  - United Arab Emirates - Dubai - Ethical AI Guidelines — Guideline, In Force
  Status verified 2026-09-09

Not found in the instruments we track (14)

Austria
Not found in the instruments we track
As an EU member state, the EU AI Act also applies here.
Côte d'Ivoire
Not found in the instruments we track
Ghana
Not found in the instruments we track
Hong Kong
Not found in the instruments we track
Iceland
Not found in the instruments we track
Iran
Not found in the instruments we track
Iraq
Not found in the instruments we track
Latvia
Not found in the instruments we track
As an EU member state, the EU AI Act also applies here.
Maldives
Not found in the instruments we track
Mexico
Not found in the instruments we track
Senegal
Not found in the instruments we track
Slovenia
Not found in the instruments we track
As an EU member state, the EU AI Act also applies here.
Tunisia
Not found in the instruments we track
Uruguay
Not found in the instruments we track

International organisations (24)

African Union
Not found in the instruments we track
APEC
Not found in the instruments we track
ASEAN
Guidance or policy only
Organizations are advised to establish human oversight, accountability, control and intervention capabilities.
- ASEAN AI Governance Guide, Compliance Checklist — Guideline, In Force
Status verified 2026-09-07
BRICS
Not found in the instruments we track
Council of Europe
Not found in the instruments we track
European Union
Yes — binding law in force
Deployers of high-risk AI systems must ensure human oversight of their operation.
- EU AI Act, Implementation Framework — Regulation, In Force (Amended)
- EU AI Act Enforcement, Compliance Checklist — Regulation, In Force
- EU Machinery Regulation 2023: AI Safety, Annex III, Section 3.2.4 — Regulation, In Force
- and 8 more
A possible change to a cited instrument is not yet verified.
Status verified 2026-09-11
G7
Not found in the instruments we track
GPAI
Not found in the instruments we track
IAEA
Guidance or policy only
It recommends determining the appropriate level of AI oversight and autonomy before deployment.
- IAEA AI Nuclear Safety Security Guidance, 5.2 — Guideline, In Force
Status verified 2026-09-09
ICAO
Not found in the instruments we track
IEEE
Not found in the instruments we track
ILO
Guidance or policy only
Final hiring, firing, task-allocation, and performance decisions must remain with human beings.
- ILO Work for a Brighter Future Report, Definitions — Policy, Adopted (from 2019-01-22)
- ILO AI and Work Observatory, Definitions — Policy, In Force
Status verified 2026-09-09
International
Yes — binding law in force
ALKS users must remain able to take over when the system requests it.
- UN Regulation on Automated Lane Keeping Systems, Key Focus Areas — Regulation, In Force (Amended)
Status verified 2026-09-08
Interpol
Guidance or policy only
Law enforcement agencies are encouraged to establish internal processes to track AI performance and assess impacts.
- Interpol Responsible AI Toolkit for Law Enforcement, Monitoring and Evaluation — Policy, In Force
Status verified 2026-09-07
ISO
Guidance or policy only
The text directs clarification of life-cycle checkpoints for human oversight.
- ISO AI System Life Cycle Processes, Key Provisions — Standard, In Force
Status verified 2026-09-07
ITU
Guidance or policy only
It calls for human-in-the-loop AI and human oversight in sensitive decision-making domains.
- ITU AI for Good Summit Report 2017, Ethical Considerations and Human Rights — Policy, Adopted (from 2017-06-09)
Status verified 2026-09-07
NATO
Guidance or policy only
AI applications must remain under human oversight, including possible human override or deactivation.
- NATO Responsible AI Principles, Definitions — Governability — Policy, In Force
Status verified 2026-09-07
OAS
Not found in the instruments we track
OECD
Guidance or policy only
AI systems must not be deployed or operated without ongoing human oversight and risk mitigation.
- OECD AI Due Diligence Guidance — Guideline, In Force
- OECD AI Principles, Section 1, 1.2(b). Respect for the rule of law, human rights — Guideline, In Force (Amended)
- OECD AI Principles, Section 1, 1.2(b) — Guideline, In Force (Amended)
Status verified 2026-09-09
UNESCO
Not found in the instruments we track
United Nations
Not found in the instruments we track
WHO
Guidance or policy only
It requires health AI to use human-in-the-loop or human-on-the-loop architecture.
- WHO Health AI Ethics Guidance, Key Provisions — Guideline, In Force
Status verified 2026-09-09
World Bank
Not found in the instruments we track
World Economic Forum
Guidance or policy only
Organizations are encouraged to create cross-functional councils to oversee AI agents throughout their lifecycle.
- WEF AI Agents Evaluation and Governance, Governance and Institutional Framework — Policy, In Force
- Global - AI Governance Framework, Implementation Framework — Policy, Adopted
Status verified 2026-09-09

Which countries require human oversight of AI?

Yes — binding law in force (12)

Argentina

Brazil

China

Finland

Hungary

Italy

Kazakhstan

Malta

Peru

Singapore

Slovakia

Vietnam

Partly — regional, framework or general (non-AI) law (6)

Estonia

India

Nigeria

Spain

United Kingdom

United States

Proposed (21)

Bahrain

Bangladesh

Chile

Colombia

Costa Rica

Croatia

Czech Republic

Ecuador

France

Indonesia

Luxembourg

Morocco

Netherlands

Norway

Panama

Philippines

Poland

Thailand

Turkey

Ukraine

Uzbekistan

Guidance or policy only (32)

Australia

Belgium

Bulgaria

Canada

Cyprus

Denmark

Egypt

Ethiopia

Germany

Greece

Ireland

Israel

Japan

Kenya

Lebanon

Lithuania

Malaysia

New Zealand

Pakistan

Portugal

Qatar

Romania

Russia

Rwanda

Saudi Arabia

Serbia

South Africa

South Korea

Sweden

Switzerland

Taiwan

United Arab Emirates

Not found in the instruments we track (14)

Austria

Côte d'Ivoire

Ghana