ต่อไปคำแนะนำต่อไป
Gender Shades and Facial Recognition Bias Audits
สังคม
คู่มือสังคม
NYC Local Law 144 requires employers and employment agencies that use automated employment decision tools (AEDTs) for hiring or promotion in New York City to get an independent bias audit each year, publish a summary of the results, and notify candidates before the tool is used.
It was one of the first laws anywhere to require bias audits of hiring algorithms, and its early record shows both what such audits can reveal and how hard they are to enforce.
The New York City Council passed Local Law 144 in 2021. It took effect on January 1, 2023, and the Department of Consumer and Worker Protection (DCWP) began enforcing it on July 5, 2023. An AEDT is a computational process derived from machine learning, statistical modeling, data analytics or AI that issues a simplified output, such as a score, classification or recommendation, used to substantially assist or replace discretionary decision-making for employment decisions. The final rules read substantially assist narrowly: for example, relying solely on the output, weighting it more than other criteria, or using it to overrule human judgment. Critics argue this lets many employers decide their tools are out of scope. Before using an AEDT, an employer must have a bias audit conducted by an independent auditor within the past year and post a summary online, including the date of the audit, the distribution date of the tool, and selection or scoring rates and impact ratios by sex, race/ethnicity and intersectional categories. Candidates and employees who are NYC residents must receive notice at least 10 business days before use, including the job qualifications and characteristics the tool assesses, and instructions for requesting an alternative process or accommodation. The law does not require the employer to provide that alternative, and it does not set a passing score: an audit showing disparities does not by itself make use unlawful, though federal and state anti-discrimination laws still apply. Penalties are $500 for a first violation and up to $1,500 for subsequent ones. Enforcement has looked weak so far. A 2024 academic study found that only a small fraction of hundreds of employers checked had posted audits, and a New York State Comptroller audit released in late 2025 criticized DCWP's complaint-driven enforcement. A common misconception is that the law bans biased tools; it mainly mandates transparency.
ความเสียหายที่เกิดจาก AI ที่เป็นหายนะและเกิดขึ้นทุกวันนั้นขึ้นอยู่กับว่าใครเข้าใจความเสี่ยงและใครสามารถดำเนินการได้
ความรู้สาธารณะและวิชาชีพเป็นตัวกำหนดว่านโยบายความปลอดภัยที่เข้มงวดจะเป็นไปได้ทางการเมืองหรือไม่
คำอธิบายที่ชัดเจนช่วยลดการจับภาพโดยการโฆษณาเกินจริง การประชาสัมพันธ์ในห้องปฏิบัติการ และการแสดงจริยธรรมที่คลุมเครือ
Local Law 144 has become a reference point for other proposals, including state bills and the Colorado AI Act, which place hiring among high-risk uses. Its early record has pushed policymakers to consider broader definitions of covered tools, clearer enforcement, and duties that go beyond disclosure. City officials and council members have discussed amendments in response to criticism, though the outcome is uncertain. For employers, the lasting lesson is that impact ratio testing, independent audits and candidate notice are becoming standard practice for AI in hiring regardless of where rules finally settle.
A retailer hiring in Manhattan uses software that ranks applicants by resume match, so it hires an independent auditor, calculates impact ratios by sex and race/ethnicity, and posts the summary on its careers page.
A staffing agency using an AI video-interview scoring tool for NYC roles sends candidates notice at least 10 business days in advance, explaining the tool will be used and how to request an alternative process or accommodation.
A company uses a tool only to schedule interviews and does not use it to score or rank candidates, so it concludes the tool likely does not substantially assist a hiring decision under the rules.
An auditor finds that fewer than 2 percent of the audit data comes from one race category, and the published summary notes that category was excluded from the impact ratio calculation as the rules allow.
การรักษาความเสี่ยงที่มีอยู่เป็นไซไฟในขณะที่สารประกอบความสามารถ
ความปลอดภัยของผลิตภัณฑ์พื้นผิวที่สับสนด้วยการจัดตำแหน่งภายใต้ความเป็นอิสระสูง
ปล่อยให้ผู้ชมที่ไม่ใช่ภาษาอังกฤษและไม่ใช่ผู้เชี่ยวชาญเหลือเพียงแหล่งข้อมูลคุณภาพต่ำ
แยกอันตรายของผลิตภัณฑ์ การใช้ในทางที่ผิด และความเสี่ยงในการสูญเสียการควบคุม/การวางแนวที่ไม่ถูกต้อง
ถามว่าหลักฐานใดที่จะเปลี่ยนมุมมองของคุณเกี่ยวกับลำดับเวลาและความรุนแรง
ชอบแหล่งที่มาหลักและการประเมินที่เป็นรูปธรรมมากกว่าคำกล่าวอ้างทางการตลาด
ระบุเส้นทางการดำเนินการเส้นทางเดียว: อาชีพ นโยบาย เงินทุน หรือทักษะ ไม่ใช่แค่ความตระหนักรู้เท่านั้น
Free newsletter
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
NYC Local Law 144 requires employers and employment agencies that use automated employment decision tools (AEDTs) for hiring or promotion in New York City to get an independent bias audit each year, publish a summary of the results, and notify candidates before the tool is used. It was one of the first laws anywhere to require bias audits of hiring algorithms, and its early record shows both what such audits can reveal and how hard they are to enforce.
DCWP adopted the rules and began enforcement on July 5, 2023.
Each category's selection rate is compared with the highest category's selection rate.
The law requires notice at least 10 business days before use.
The law mandates transparency rather than a pass/fail standard, though other anti-discrimination laws still apply.
Scoring rates use the sample median as the cutoff, then are compared across categories.
เรียนรู้ต่อไป
คำแนะนำเพิ่มเติมที่เลือกสำหรับหัวข้อนี้
ต่อไปคำแนะนำต่อไป
Gender Shades and Facial Recognition Bias Audits
สังคม