What does NYC Local Law 144 (AEDT) say about AI in banking?

What are the key points of NYC Local Law 144 (AEDT)?

• Administrative Code § 20-870 defines an automated employment decision tool as any computational process, derived from machine learning, statistical modeling, data analytics or artificial intelligence, that issues a simplified output (a score, classification or recommendation) used to substantially assist or replace discretionary decision making for employment decisions that impact natural persons; 'employment decision' means to screen candidates for employment or employees for promotion within the city.
• § 20-871(a): unlawful to use an AEDT to screen a candidate or employee unless (1) it has been the subject of a bias audit conducted no more than one year before use, and (2) a summary of the most recent bias audit and the tool's distribution date have been made publicly available on the employer's or employment agency's website before use.
• § 20-871(b): notice to each candidate or employee who resides in the city, no less than 10 business days before use, that an AEDT will be used, allowing a candidate to request an alternative selection process or accommodation, and of the job qualifications and characteristics the tool will use; information on data type, source and retention must be available on written request within 30 days.
• § 20-872: civil penalty of not more than $500 for a first violation and each additional violation on the same day, and $500–$1,500 for each subsequent violation; each day an AEDT is used in violation is a separate violation, and failure to give any required notice is a separate violation; proceedings are returnable to the Office of Administrative Trials and Hearings or a designated city agency.
• §§ 20-873 and 20-874: the corporation counsel, on behalf of the department, may bring an action to compel compliance; the subchapter does not limit a candidate's or employee's right to sue or the Commission on Human Rights' authority under Title 8.
• 6 RCNY § 5-300: an AEDT 'substantially assists or replaces discretionary decision making' only if it relies solely on a simplified output, uses a simplified output weighted more than any other criterion, or uses it to overrule conclusions from other factors including human decision-making.
• 6 RCNY §§ 5-301 to 5-302: a bias audit must calculate selection rates (or scoring rates) and impact ratios by sex, race/ethnicity and intersectional categories, indicate the number of individuals in unknown categories, may exclude a category under 2% of the data with a stated justification, must use historical data (or test data where historical data is insufficient), and must be conducted by an independent auditor with no role in the tool and no employment or financial relationship with the employer or vendor.
• 6 RCNY §§ 5-303 to 5-304: results and distribution date must be posted before use on the employment section of the website and kept posted at least 6 months after the latest use; notice must be given at least 10 business days before use by website, job posting, mail or e-mail (or a written policy for employees); the rule does not require an alternative selection process.

What did NYC Local Law 144 (AEDT) change for banks?

Local Law 144 was an early U.S. law requiring independent bias audits and public disclosure for AI hiring tools, and DCWP's 2023 rule narrowed the definition of covered tools and prescribed the impact-ratio method. For banks it means any résumé-ranking, video-scoring or promotion-scoring tool used for New York City roles needs an audit within the past year, a posted summary and advance notice. The State Comptroller's December 2025 audit of DCWP's enforcement (Report 2024-N-6) suggests a more proactive enforcement posture may follow.

What does NYC Local Law 144 require of banks that use AI in hiring?

NYC Local Law 144 of 2021 requires an employer or employment agency, including a bank, that uses an automated employment decision tool to screen candidates or employees in New York City to do three things before use. It must have the tool bias-audited by an independent auditor no more than one year before use; publish the date of the audit, a summary of the results and the tool's distribution date on the employment section of its website; and notify each candidate or employee who resides in the city at least 10 business days in advance, including how to request an alternative process or accommodation and what job qualifications the tool will assess. DCWP's rule specifies the audit: selection or scoring rates and impact ratios for sex, race/ethnicity and intersectional categories, with unknown-category counts. Penalties run $500 to $1,500 per violation with each day a separate violation, and DCWP has enforced since July 5, 2023.

Local Law 144 is a transparency-and-audit regime, not a ban: a bank may use an AEDT so long as the audit is current, the results are posted and notices are given. DCWP's rule shaped it in two ways. It narrowed the definition so that a tool counts only if it relies solely on a simplified output, weights it more than any other criterion, or uses it to overrule other conclusions; a tool that merely feeds a human reviewer who weighs many factors may fall outside, though a ranking or recommendation that drives who is interviewed will usually be inside. And it fixed the audit as an impact-ratio calculation, drawn from the EEOC's Uniform Guidelines, across sex, race/ethnicity and intersectional categories.

The law's scope is geographic and functional. It applies to screening in the city and to candidates and employees who reside in the city, and an employer relying on a vendor's audit may do so only if it provided its own historical data to the auditor or has never used the tool (6 RCNY § 5-302(a)). A bank that hires nationally should confirm which tools touch New York City applicants; a single AI résumé screener used across locations can bring New York City postings within the law. The law does not reach AI in credit, fraud or customer service, which are governed by federal and state financial rules.

Enforcement has been modest but is under pressure. The Comptroller's audit of DCWP's work from July 2023 to June 2025 found weak complaint intake, no check of whether the intake process worked despite only two complaints, and a compliance review of 32 companies that found one issue where the auditors found at least 17. DCWP generally agreed with the recommendations, which include enforcement not dependent on complaints and use of the Office of Technology and Innovation's expertise. Banks should not read past enforcement as a safe harbor: penalties accrue daily, private claims and Commission on Human Rights action remain available under § 20-874, and posted audit summaries are public.

WHAT THIS MEANS IN PRACTICE

• Inventory HR tools used for New York City roles, and for each decide with documentation whether it is an AEDT under 6 RCNY § 5-300; keep the reasoning for tools judged out of scope.
• Obtain or commission an independent bias audit dated within 12 months of use, check the auditor's independence under § 5-300, and diarize renewal before the anniversary.
• If relying on a vendor's audit, confirm it used historical data and that your own historical data was supplied, or that you have never used the tool (§ 5-302(a)); otherwise commission your own audit or rely on test data with the stated explanation.
• Post the audit date, results summary and distribution date on the employment section of your website before use, and keep them posted at least 6 months after the tool's last use (§ 5-303).
• Send the § 20-871(b) notice at least 10 business days before first use by website, posting, mail or e-mail, and include how to request an alternative process or accommodation.
• Treat each day of use and each missing notice as a separate violation when sizing exposure, and align the program with Illinois HB 3773 and California's employment ADS regulations for multistate hiring.