# AI Audit Requirements: Which Laws Require an AI Audit, Who Audits, What Is Checked, and When

Source: https://superintelligencecompliance.ai/guides/ai-audit  
Publisher: ZSKFL Management  
Last updated: 2026-11-16. Not legal advice.

> NYC requires bias audits of hiring and promotion tools; Illinois, audits of large frontier developers from 2028. No US statute here requires one of every firm.

## At a glance

- As of November 15, 2026, no US federal or state statute covered on this site requires every company to audit its AI.
- NYC Local Law 144 bars employers and employment agencies from using an automated employment decision tool to screen candidates or employees for hiring or promotion in the city unless the tool had an independent bias audit no more than one year before use and a summary of the results is posted.
- Illinois’s AI Safety Measures Act requires a large frontier developer, from January 1, 2028 or 90 days after it first qualifies, whichever is later, to annually retain a third party to perform an independent audit of compliance with Section 10. California SB 53 and New York’s amended RAISE Act contain no audit requirement.
- Under the EU AI Act, providers of high-risk systems in Annex III points 2 to 8 assess conformity themselves under Annex VI, with no notified body, from December 2, 2027. Biometric systems in Annex III point 1, from the same date, and Annex I Section A products, from August 2, 2028, may need a notified body. ISO/IEC 42001 certification is voluntary.

## Which laws require an AI audit?

As of November 15, 2026, two US laws covered on this site require an independent AI audit by name: NYC Local Law 144, for hiring and promotion tools, and Illinois SB 315, for large frontier developers from 2028. The EU AI Act requires a conformity assessment of high-risk systems from December 2, 2027 or August 2, 2028, and for most Annex III systems the provider performs it itself. Insurance guidance expects internal audit to review AI risk management. No law covered here requires a general AI audit of every company.

“AI audit” names several different checks: a statistical test of one tool, a compliance review of a developer’s safety program, a product conformity assessment, a regulator’s evaluation of a model, a voluntary certification, and internal audit inside a regulated firm. An impact assessment is a separate instrument, covered in the [AI impact assessment guide](https://superintelligencecompliance.ai/guides/ai-impact-assessment).

## What kinds of AI audit does each law require?

**AI audits and audit-like reviews by law, as of November 15, 2026**

| Audit type | Law or rule | Who must be audited | Who audits | What is checked | Binding? | Date |
|---|---|---|---|---|---|---|
| Bias audit | NYC Local Law 144 (Admin. Code 20-870 to 20-874; 6 RCNY 5-300 to 5-304) | Employers and employment agencies using an automated employment decision tool in the city | An independent auditor; no DCWP approval needed | Disparate impact on sex, race or ethnicity, and intersectional categories, by selection or scoring rates and impact ratios | Binding | No more than one year before use; enforced since July 5, 2023 |
| Frontier developer compliance audit | Illinois AI Safety Measures Act (SB 315, Public Act 104-0538), Section 10(d) | Large frontier developers | A third party with demonstrated competence, using generally accepted auditing standards and best practices | Compliance with Section 10, including the developer’s internal controls | Binding | Annually from January 1, 2028, or 90 days after first qualifying if later |
| No audit; third-party assessment described in a framework | California SB 53 (Bus. & Prof. Code 22757.12(a)(5)); New York RAISE Act as amended (GBL 1421(1)(e)) | Large frontier developers | The developer decides whether and how to use third parties | The framework must describe using third parties to assess catastrophic risks and the effectiveness of mitigations | Binding framework duty; no audit duty | California since January 1, 2026; New York from January 1, 2027 |
| EU conformity assessment, internal control | EU AI Act Article 43(2), Annex VI | Providers of high-risk systems in Annex III points 2 to 8 | The provider; no notified body | Quality management system against Article 17, technical documentation against the Section 2 requirements, and consistency of design, development, and post-market monitoring with that documentation | Binding | December 2, 2027 |
| EU conformity assessment, notified body | EU AI Act Article 43(1), Annex VII | Providers of biometric systems in Annex III point 1 that choose this route or must use it | A notified body independent of the provider; the market surveillance authority for systems put into service by law enforcement, immigration or asylum authorities, or EU institutions, bodies, offices, or agencies | Quality management system and technical documentation; further evidence or tests on request, and the notified body’s own tests where it is not satisfied with the provider’s; periodic surveillance audits of the quality management system, with an audit report to the provider | Binding | December 2, 2027 |
| EU sectoral conformity assessment | EU AI Act Article 43(3) as amended | Providers of high-risk systems covered by Annex I Section A product law | The procedure the product law sets, which may involve a notified body | The Section 2 requirements and the Article 17 quality management system, inside the product assessment | Binding | August 2, 2028; notified bodies apply for AI Act designation by January 28, 2028 |
| Regulator evaluation of a general-purpose AI model | EU AI Act Article 92; Implementing Regulation (EU) 2026/1755 | Providers of general-purpose AI models; systemic-risk investigations reach models with systemic risk | The AI Office, after consulting the Board, or independent experts the Commission appoints | Compliance where information gathered under Article 91 is insufficient, or systemic risks at Union level | Binding power of the regulator | Since August 2, 2026; the implementing regulation entered into force August 10, 2026 |
| Management system certification | ISO/IEC 42001:2023; ISO/IEC 42006:2025 for certification bodies | Any organization that chooses it | An independent certification body, which may be accredited by a national accreditation body | The AI management system against ISO/IEC 42001 | Voluntary | 42001: December 2023; 42006: July 2025 |
| Insurer internal audit and examination | NAIC Model Bulletin; NYDFS Circular Letter No. 7; Colorado Regulation 10-1-1 | Insurers, in adopting states (NAIC), authorized in New York (Circular Letter No. 7), or covered by the Colorado rule | The insurer’s internal audit function and department examiners (NAIC, NYDFS); Colorado’s rule names no auditor and requires an annual compliance report signed by an officer (Section 6) | Effectiveness of the AI program, validation, documentation, data, bias, and board reporting | Guidance (NAIC, NYDFS); binding governance framework rule (Colorado insurers) | NAIC: December 4, 2023; Circular Letter No. 7: July 11, 2024; Colorado amended rule: October 15, 2025 |
| Bank model risk review | SR 26-2, OCC Bulletin 2026-13, FDIC FIL-15-2026 | Banking organizations; most relevant above $30 billion in total assets; generative and agentic AI models out of scope | Individuals performing effective challenge, with the appropriate expertise, sufficient independence to maintain objectivity, and the organizational standing and influence to effect change; internal audit where it is part of model risk management | Whether model risk management practices are rigorous and effective and whether related policies are implemented | Guidance | April 17, 2026 |
| Federal agency independent review | OMB Memorandum M-25-21 | Federal agencies, for high-impact AI use cases | An independent reviewer within the agency who was not involved in development | The AI impact assessment before deployment | Binding on agencies only | April 3, 2025 |

## What does a NYC Local Law 144 bias audit require?

A bias audit is an impartial evaluation by an independent auditor that includes, but is not limited to, testing the tool’s disparate impact on the sex and race or ethnicity categories employers report to the EEOC on the EEO-1 Component 1 report (Admin. Code 20-870; 6 RCNY 5-300). It must use the tool’s historical data, which may be pooled across employers, but an employer may rely on a pooled audit only if it gave the auditor its own historical data or has never used the tool. Test data is allowed only if insufficient historical data is available for a statistically significant audit (5-302). No result fails the audit, and anti-discrimination laws still apply.

According to DCWP’s FAQ, the vendor that built the tool is not responsible for the audit. A vendor may commission an independent audit, but the employer or agency must make sure an audit was done before use. The [NYC Local Law 144 guide](https://superintelligencecompliance.ai/guides/nyc-local-law-144) covers data, posting, notice, and penalties.

Of the laws covered on this site, Local Law 144 is the one that requires a bias audit by name. Illinois’s HB 3773 has no audit duty in the statute. Colorado SB 26-189, which applies to consequential decisions made on or after January 1, 2027, contains no risk management program or impact assessment duty, though the 2024 law it replaced required both. Texas HB 149 requires no impact assessment, risk management program, bias audit, or general disclosure from private companies.

## Must frontier AI developers get an independent audit?

In Illinois, large frontier developers must, from 2028. California SB 53 and New York’s amended RAISE Act contain no audit requirement.

Under Section 10(d) of the Illinois Act, beginning on January 1, 2028 or 90 days after a developer first qualifies as a large frontier developer, whichever is later, a large frontier developer must annually retain a third party to perform an independent audit of compliance with the requirements of Section 10, which also holds the frontier AI framework and transparency report duties. A large frontier developer is one that, with its affiliates, had more than $500 million in annual gross revenues in the preceding calendar year.

- **Auditor.** The third party must conduct audits consistent with generally accepted auditing standards and best practices and possess demonstrated competence to perform the audit, including experience employing or contracting with individuals who possess technical expertise in the safety of frontier models.
- **Report.** It must state whether the developer substantially complied with Section 10, describe any material deviations with their rationale and recommendations, assess the developer’s internal controls in detail, including its designation and empowerment of senior personnel responsible for implementation, list the audit personnel, give the auditor’s conflict-of-interest procedures and any conflicts, describe the methodology and information reviewed, and carry the lead auditor’s signature certifying the results.
- **Publication.** Within 30 days after receiving the report, the developer must conspicuously publish on its website a high-level summary of the findings and a copy of the report with appropriate redactions, transmit the redacted report to the Illinois Emergency Management Agency and Office of Homeland Security and the Attorney General, and keep an unredacted copy for as long as a frontier model is deployed plus 5 years.
Failing to have a third party perform the required audit is among the failures for which only the Attorney General may recover civil penalties under Section 25, of up to $1,000,000 for a first violation and up to $3,000,000 for each subsequent violation. The [Illinois AI laws guide](https://superintelligencecompliance.ai/guides/illinois-ai-laws) covers the rest of the Act, including Section 17 designations of substantially equivalent federal standards for third-party audits.

California’s SB 53, as chaptered on September 29, 2025 (Chapter 138, Statutes of 2025), contains no audit requirement. The Assembly amendments of July 17 and September 2, 2025 would have required large developers to retain an independent third-party auditor annually beginning January 1, 2030, and the September 5 amendment removed that provision. New York’s RAISE Act, as amended by Chapter 96 of 2026 and effective January 1, 2027, contains no audit requirement in sections 1420 to 1429 as read on November 15, 2026. Both instead require a large frontier developer’s framework to describe using third parties to assess the potential for catastrophic risks and the effectiveness of mitigations (Bus. & Prof. Code 22757.12(a)(5); GBL 1421(1)(e)), and its transparency report to state the extent to which third-party evaluators were involved in its catastrophic risk assessments. The [New York AI laws guide](https://superintelligencecompliance.ai/guides/new-york-ai-laws) covers RAISE.

## Is an EU AI Act conformity assessment an audit?

It is the closest EU equivalent, and for most Annex III systems the provider performs it on itself. On the Annex VII route, a notified body also audits the quality management system periodically and gives the provider an audit report (point 5). From December 2, 2027 (Annex III) or August 2, 2028 (Annex I Section A), a provider must complete the Article 43 conformity assessment before placing a high-risk system on the market or putting it into service, and repeat it after any substantial modification (Article 43(4)).

- **Annex III points 2 to 8**: internal control under Annex VI, which does not provide for the involvement of a notified body (Article 43(2)).
- **Annex III point 1 (biometrics):** where the provider has applied harmonised standards or, where applicable, common specifications, it may choose Annex VI or the Annex VII notified body route. It must follow Annex VII where harmonised standards do not exist and common specifications are not available, where it has not applied, or has applied only part of, the harmonised standard, where it has not applied existing common specifications, or, for the restricted part only, where a harmonised standard has been published with a restriction (Article 43(1)). As of November 15, 2026, no harmonised standard had been cited for the requirements in Articles 9 to 15.
- **Annex I Section A products:** the sectoral procedure applies, the Section 2 requirements are checked inside it, and the Article 17 quality management system is also assessed (Article 43(3) as amended). The [EU high-risk AI systems guide](https://superintelligencecompliance.ai/guides/eu-ai-act-high-risk-ai-systems) covers this route.
Separately, under Article 92(1), applicable since August 2, 2026, the AI Office, after consulting the Board, may conduct evaluations of a general-purpose AI model to assess the provider’s compliance where the information gathered under Article 91 is insufficient, or to investigate systemic risks at Union level of models with systemic risk. The Commission may appoint independent experts to carry out evaluations on its behalf.

## Does ISO/IEC 42001 certification count as an AI audit?

It is an audit, but no law covered on this site requires it. ISO/IEC 42001:2023 is a paid, voluntary management system standard that an independent certification body can audit a company against and certify. ISO/IEC 42006:2025, published July 2025, sets additional requirements, on top of ISO/IEC 17021-1, for bodies that audit and certify AI management systems against it. A certificate does not test a tool for disparate impact, replace an EU conformity assessment, or meet Illinois’s audit duty. The [AI governance frameworks guide](https://superintelligencecompliance.ai/guides/ai-governance-frameworks) compares it with the NIST AI RMF, which NIST audits no one against.

## What do insurance and banking regulators expect from internal audit and examinations?

- **NAIC Model Bulletin.** In adopting states (27 states plus the District of Columbia and Puerto Rico on the NAIC’s list as of November 3, 2026), the written AIS Program should address governance, risk management controls, and internal audit functions. In an investigation or market conduct action, an insurer can expect to be asked for documentation related to validation, testing, and auditing, including evaluation of model drift.
- **NYDFS Circular Letter No. 7.** The letter notes that 11 NYCRR 89.16 requires an insurer to have an internal audit function, and says insurers should ensure that function is appropriately engaged with their use of external consumer data and AI systems, consistent with the financial, operational, and compliance risk. The Department may audit and examine an insurer’s use of those data and systems in regular or targeted examinations under Insurance Law 309, or through a request for special report under section 308.
- **Colorado Regulation 10-1-1.** For the insurers it covers, the governance framework must be designed to determine whether the use of external consumer data, and of algorithms and predictive models that use it, potentially results in unfair discrimination with respect to race, and to remediate unfair discrimination detected through quantitative testing requirements the Division establishes (Section 5.A). The [insurance guide](https://superintelligencecompliance.ai/guides/ai-compliance-insurance) tracks those requirements.
- **Banks.** Under SR 26-2, OCC Bulletin 2026-13, and FDIC FIL-15-2026, where internal audit is part of a bank’s model risk management, it would generally not duplicate model development or validation. Its role is generally to evaluate whether the practices are rigorous and effective and whether related policies are implemented accordingly. Generative and agentic AI models are not within the guidance’s scope (footnote 3), and non-compliance with it will not by itself result in supervisory criticism.

## What makes an AI auditor independent?

Each rule that names an auditor sets its own test, and meeting one does not meet the others.

- **NYC Local Law 144.** An auditor is not independent if it is or was involved in using, developing, or distributing the tool, or if at any point during the audit it has an employment relationship with, or a direct financial interest or material indirect financial interest in, an employer or agency that seeks to use or continue to use the tool or the vendor that developed or distributes it. The State Comptroller counted three of 22 employers’ audits as potentially not independent, including two where an independent auditor wrote the report but the vendor performed the testing.
- **Illinois SB 315.** A large frontier developer must not retain a third party if either one has a financial interest in the other, and may pay the auditor but must not condition any payment, or its amount, on the results.
- **EU notified bodies.** A notified body must be independent of the provider whose system it assesses, of any other operator with an economic interest in that system, and of the provider’s competitors, and the body, its top-level management, and its assessment staff must not be directly involved in the design, development, marketing, or use of high-risk AI systems (Article 31(4) and (5)).
- **EU independent experts.** The Commission takes into account shared ownership, governance, management, personnel, or resources, any previous appointment to evaluate on its behalf, and contractual relationships with the provider concerned or any other provider over at least the 12 months before the evaluation (Implementing Regulation 2026/1755, Article 3(1)).

## What does an AI auditor check?

The rule sets the scope. Circular Letter No. 7 presents its items as among those insurer internal audit may include.

- **Outcomes by group:** selection or scoring rates and impact ratios (Local Law 144); unfair discrimination with respect to race, under quantitative testing requirements the Division establishes (Colorado Regulation 10-1-1)
- **Data:** historical or permitted test data (Local Law 144); the accuracy, reliability, and integrity of data, and potential biases that may result in unfair or unlawful discrimination (Circular Letter No. 7)
- **The program:** the quality management system against Article 17 (EU Annex VI and VII); whether policies are in place and followed (Circular Letter No. 7)
- **Documentation:** technical documentation against the Section 2 requirements (EU Annex VI and VII)
- **Testing:** the provider’s tests, and the notified body’s own where it is not satisfied with them (EU Annex VII)
- **Controls and compliance:** internal controls, senior personnel, and substantial compliance with Section 10 (Illinois)

## How do you prepare for an AI audit?

- List each AI system with the audit type, if any, that each jurisdiction where it is used attaches to it, and the date each first applies
- For each automated employment decision tool used for hiring or promotion in New York City, confirm that an independent bias audit was done no more than one year before use, that the auditor, not the vendor, performed the testing, and that the summary is posted
- Test each auditor against the independence definition in the rule that requires the audit, and record the conflict check
- If you may become a large frontier developer under the Illinois Act, plan an annual third-party audit from January 1, 2028, or 90 days after you first qualify if later, with payment terms that do not depend on the results
- For each EU high-risk system you provide, identify the Article 43 route, and check whether the harmonised standards or common specifications you rely on have been cited
- At an insurer, give internal audit a scope that reaches AI systems, as the NAIC bulletin and Circular Letter No. 7 expect
How this fits the other guides: the [AI governance policy guide](https://superintelligencecompliance.ai/guides/ai-governance-policy) covers the policy an auditor reviews.

This guide is general information as of November 15, 2026, not legal advice. Confirm current requirements with the primary source, your regulator, or counsel before relying on an audit to meet a legal duty.

## Sources

- [Int. No. 1894-A, Local Law 144 of 2021 (Administrative Code 20-870 to 20-874), legislation text and history](https://legistar.council.nyc.gov/LegislationDetail.aspx?ID=4344524&GUID=B051915D-A9AC-451E-81F8-6596032FA3F9&Options=ID%7CText%7C&Search=), New York City Council, accessed October 26, 2026.
- [Notice of Adoption of Final Rule, Automated Employment Decision Tools (6 RCNY 5-300 to 5-304) (PDF)](https://rules.cityofnewyork.us/wp-content/uploads/2023/04/DCWP-NOA-for-Use-of-Automated-Employment-Decisionmaking-Tools-2.pdf), NYC Department of Consumer and Worker Protection, April 2023.
- [Automated Employment Decision Tools: Frequently Asked Questions (PDF)](https://www.nyc.gov/assets/dca/downloads/pdf/about/DCWP-AEDT-FAQ.pdf), NYC Department of Consumer and Worker Protection, June 29, 2023.
- [Enforcement of Local Law 144: Automated Employment Decision Tools, audit report 2024-N-6 (PDF)](https://www.osc.ny.gov/files/state-agencies/audits/pdf/sga-2026-24n6.pdf), Office of the New York State Comptroller, December 2, 2025.
- [Public Act 104-0538 (SB 315), Artificial Intelligence Safety Measures Act (PDF)](https://www.ilga.gov/Documents/legislation/publicacts/104/PDF/104-0538.pdf), Illinois General Assembly, approved July 6, 2026, effective January 1, 2027, Sections 5, 10, 17, and 25.
- [SB 315, enrolled text](https://www.ilga.gov/ftp/legislation/104/SB/10400SB0315enr.htm), Illinois General Assembly, accessed November 15, 2026.
- [Public Act 103-0804 (HB 3773)](https://www.ilga.gov/Legislation/publicacts/view/103-0804), Illinois General Assembly, approved August 9, 2024, effective January 1, 2026.
- [California Legislative Information bulk data, pubinfo_2025.zip: SB 53 chaptered text (Chapter 138, Statutes of 2025, September 29, 2025) and the Assembly amendments of July 17, September 2, and September 5, 2025; Business and Professions Code sections 22757.10 to 22757.16](https://downloads.leginfo.legislature.ca.gov/pubinfo_2025.zip), California Legislature, file dated November 9, 2026, accessed November 15, 2026.
- [General Business Law Article 44-B (RAISE Act), sections 1420 to 1429](https://legislation.nysenate.gov/pdf/laws/GBSA44-B), New York State Senate, accessed November 15, 2026.
- [A.9449 (Chapter 96 of 2026)](https://legislation.nysenate.gov/pdf/bills/2025/A9449), New York State Senate, accessed November 15, 2026.
- [Regulation (EU) 2024/1689 (Artificial Intelligence Act), consolidated text as of July 27, 2026, as amended by Regulation (EU) 2026/1744](https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02024R1689-20260727), EUR-Lex, Articles 16, 31, 43, 92, and 113 and Annexes VI and VII, accessed November 15, 2026.
- [Commission Implementing Regulation (EU) 2026/1755 of 20 July 2026 on detailed arrangements for the conduct of certain proceedings by the Commission pursuant to Regulation (EU) 2024/1689](https://eur-lex.europa.eu/eli/reg_impl/2026/1755/oj/eng), Official Journal of the European Union, L series, July 21, 2026.
- [Commission Implementing Decision (EU) 2026/2274 of 12 October 2026 on the harmonised standards for high-risk AI systems drafted in support of Regulation (EU) 2024/1689](https://eur-lex.europa.eu/eli/dec_impl/2026/2274/oj/eng), Official Journal of the European Union, L series, October 14, 2026. Its one reference, EN 18286:2026, covers the quality management system; as of November 15, 2026, it was the only AI Act harmonised standard reference published.
- [ISO/IEC 42001:2023, Information technology, Artificial intelligence, Management system](https://www.iso.org/standard/42001), International Organization for Standardization, December 2023.
- [ISO/IEC 42006:2025, Information technology, Artificial intelligence, Requirements for bodies providing audit and certification of artificial intelligence management systems](https://www.iso.org/standard/42006), International Organization for Standardization, July 2025.
- [NAIC Model Bulletin: Use of Artificial Intelligence Systems by Insurers](https://content.naic.org/sites/default/files/inline-files/2023-12-4%20Model%20Bulletin_Adopted_0.pdf), National Association of Insurance Commissioners, December 4, 2023.
- [Implementation of NAIC Model Bulletin: Use of Artificial Intelligence Systems by Insurers (status as of November 3, 2026)](https://content.naic.org/sites/default/files/legal-adoption-map-ai-model-bulletin.pdf), National Association of Insurance Commissioners, November 2026.
- [Insurance Circular Letter No. 7 (2024): Use of Artificial Intelligence Systems and External Consumer Data and Information Sources in Insurance Underwriting and Pricing](https://www.dfs.ny.gov/industry-guidance/circular-letters/cl2024-07), New York State Department of Financial Services, July 11, 2024.
- [Amended Regulation 10-1-1 (3 CCR 702-10), Governance and Risk Management Framework Requirements for Life Insurers’, Private Passenger Automobile Insurers’, and Health Benefit Plan Insurers’ Use of External Consumer Data and Information Sources, Algorithms, and Predictive Models](https://drive.google.com/file/d/1gWvUT_zEn6npupumvYB4Sq_hvh2OZO8y/view), Colorado Division of Insurance, effective October 15, 2025, linked from the Division’s [notice of adoption](https://doi.colorado.gov/announcements/notice-of-adoption-amended-regulation-10-1-1-governance-and-risk-management-framework) of August 20, 2025.
- [SR 26-2 Attachment: Supervisory Guidance on Model Risk Management (PDF)](https://www.federalreserve.gov/supervisionreg/srletters/SR2602a1.pdf), Federal Reserve Board, OCC, and FDIC, April 17, 2026.
- [OCC Bulletin 2026-13, Model Risk Management: Revised Guidance](https://www.occ.gov/news-issuances/bulletins/2026/bulletin-2026-13.html), Office of the Comptroller of the Currency, April 17, 2026.
- [FIL-15-2026, Agencies Revise the Interagency Model Risk Management Guidance](https://www.fdic.gov/news/financial-institution-letters/2026/agencies-revise-interagency-model-risk-management-guidance), Federal Deposit Insurance Corporation, April 17, 2026.
- [OMB Memorandum M-25-21, Accelerating Federal Use of AI through Innovation, Governance, and Public Trust](https://www.whitehouse.gov/wp-content/uploads/2025/02/M-25-21-Accelerating-Federal-Use-of-AI-through-Innovation-Governance-and-Public-Trust.pdf), Office of Management and Budget, April 3, 2025.
- [SB 26-189, Session Laws 2026, Chapter 131 (enrolled act)](https://leg.colorado.gov/laws/session-laws/SB26-189/131/download), Colorado General Assembly, approved May 14, 2026.
- [H.B. No. 149 (enrolled), Texas Responsible Artificial Intelligence Governance Act](https://capitol.texas.gov/tlodocs/89R/billtext/html/HB00149F.htm), Texas Legislature, 2025.

## Common questions

**Is an AI audit legally required?**

As of November 15, 2026, no US federal or state statute covered on this site requires every company to audit its AI. NYC Local Law 144 bars employers and employment agencies from using an automated employment decision tool to screen candidates or employees for hiring or promotion in the city unless the tool had an independent bias audit no more than one year before use and a summary of the results is posted. Illinois's AI Safety Measures Act requires a large frontier developer, from January 1, 2028 or 90 days after it first qualifies, whichever is later, to annually retain a third party to perform an independent audit of compliance with Section 10.

**Do California SB 53 or New York's RAISE Act require a third-party audit?**

No. California SB 53, as chaptered on September 29, 2025, and New York's RAISE Act, as amended by Chapter 96 of 2026 and effective January 1, 2027, contain no audit requirement. Both require a large frontier developer's framework to describe using third parties to assess the potential for catastrophic risks and the effectiveness of mitigations, and its transparency report to state the extent to which third-party evaluators were involved in its catastrophic risk assessments.

**Does the EU AI Act require a third-party audit of high-risk AI?**

Not for most Annex III systems: from December 2, 2027, providers of high-risk systems in Annex III points 2 to 8 follow internal control under Annex VI, which does not provide for the involvement of a notified body. Providers of biometric systems in Annex III point 1 may choose Annex VI or the Annex VII notified body route where they have applied harmonised standards or, where applicable, common specifications, and must use Annex VII in the cases Article 43(1) lists, including where harmonised standards do not exist and common specifications are not available. Systems covered by Annex I Section A product law follow that law's conformity assessment procedure from August 2, 2028.

