WHAT THIS BILL REGULATES · 4 REQUIREMENT TYPES
How Is This Bill Enforced
Verbatim statutory text on the left; plain-language analysis and a per-section checklist on the right. Numbered markers cross-link to the matching checklist row.
(1)(a)–(c) "Automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a)" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisionsHousing decision"Housing decision" means to screen applicants for housing.RPL § 227-g(1)(c) that impact natural persons. "Automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a)" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data. (b) "Disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b)" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) shall include, but not be limited to, testing of the extent to which use of an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) shall differentiate between applicants who were selected and applicants who were not selected by the tool. (c) "Housing decisionHousing decision"Housing decision" means to screen applicants for housing.RPL § 227-g(1)(c)" means to screen applicants for housing.
(2)(a) 1 It shall be unlawful for a landlord to implement or use an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) that fails to comply with the following provisions: (a) No less than annually, a disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) shall be conducted to assess the actual impact of any automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) used by any landlord to select applicants for housing within the state. Such disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) shall be provided to the landlord.
(2)(b) 2 A summary of the most recent disparate impact analysis of such tool as well as the distribution date of the tool to which the analysis applies shall be made publicly available on the website of the landlord prior to the implementation or use of such tool. Such summary shall also be made accessible through any listing for housing on a digital platform for which the landlord intends to use an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) to screen applicants for housing.
(3)(a)–(b) 3 3. (a) Any landlord that uses an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) to screen applicants for housing shall notify each such applicant of the following: (i) That an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) will be used in connection with the assessment or evaluation of such applicant; (ii) The characteristics that such automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) will use in the assessment of such applicant; (iii) Information about the type of data collected for such automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a), the source of such data, and the landlord's data retention policy; and (iv) If an application for housing is denied through use of the automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a), the reason for such denial. (b) The notice required by paragraph (a) of this subdivision shall be made no less than twenty-four hours before the use of such automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) and shall allow such applicant to request an alternative selection process or accommodation.
(4) 4. The attorney general may initiate an investigation if a preponderance of the evidence, including the summary of the most recent disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) establishes a suspicion of a violation. The attorney general may also initiate in any court of competent jurisdiction any action or proceeding that may be appropriate or necessary for correction of any violation issued pursuant to this section, including mandating compliance with the provisions of this section or such other relief as may be appropriate.
This section regulates the use of automated decision tools — AI systems that screen applicants for housing — by landlords. It imposes three core obligations: an annual independent disparate impact analysis (bias audit), public disclosure of the audit summary, and pre-screening notice to each applicant with details about the tool's characteristics, data practices, and denial reasons. Applicants must receive at least 24 hours' advance notice before the tool is used and must be offered the right to request an alternative selection process. Enforcement is through the Attorney General, who may investigate upon a preponderance of evidence suggesting a violation and may bring court actions for compliance or other relief.
(1)(a)–(d) "Pricing algorithmPricing algorithm"Pricing algorithm" shall mean any computational process, including a computational process derived from machine learning or other artificial intelligence techniques, that processes data to recommend or set a price or commercial term that is in or affecting interstate or foreign commerce.GBL § 344(1)(a)" shall mean any computational process, including a computational process derived from machine learning or other artificial intelligence techniques, that processes data to recommend or set a price or commercial term that is in or affecting interstate or foreign commerce. (b) "CoordinatorCoordinator"Coordinator" shall mean any operator of a software or data analytics service that performs a coordinating function for any landlord, including a landlord performing a coordinating function for such landlord's own benefit.GBL § 344(1)(b)" shall mean any operator of a software or data analytics service that performs a coordinating function for any landlord, including a landlord performing a coordinating function for such landlord's own benefit. (c) "Nonpublic competitor dataNonpublic competitor data"Nonpublic competitor data": (i) shall mean nonpublic data that is derived from or otherwise provided by another person that competes in the same market as a person, or a related market; and (ii) does not include information distributed, reported, or otherwise communicated in a way that does not reveal any underlying data from a competitor, such as narrative industry reports, news reports, business commentaries, or generalized industry survey results.GBL § 344(1)(c)": (i) shall mean nonpublic dataNonpublic data"Nonpublic data" shall mean information that is not widely available or easily accessible to the public, including information about actual rent prices, occupancy rates, lease start and end dates, and similar data, regardless of whether such data is attributable to a specific competitor or anonymized.GBL § 344(1)(d) that is derived from or otherwise provided by another person that competes in the same market as a person, or a related market; and (ii) does not include information distributed, reported, or otherwise communicated in a way that does not reveal any underlying data from a competitor, such as narrative industry reports, news reports, business commentaries, or generalized industry survey results. (d) "Nonpublic dataNonpublic data"Nonpublic data" shall mean information that is not widely available or easily accessible to the public, including information about actual rent prices, occupancy rates, lease start and end dates, and similar data, regardless of whether such data is attributable to a specific competitor or anonymized.GBL § 344(1)(d)" shall mean information that is not widely available or easily accessible to the public, including information about actual rent prices, occupancy rates, lease start and end dates, and similar data, regardless of whether such data is attributable to a specific competitor or anonymized.
(2)(a) 4 In setting the amount of rent to be charged to a tenant for the occupancy of a residential premises, including determining any change in the amount of rent to be charged for the renewed occupancy of a residential premises, a landlord shall not employ, use, or rely upon, or cause another person to employ, use, or rely upon, an algorithmic pricing that uses, incorporates, or was trained with nonpublic competitor dataNonpublic competitor data"Nonpublic competitor data": (i) shall mean nonpublic data that is derived from or otherwise provided by another person that competes in the same market as a person, or a related market; and (ii) does not include information distributed, reported, or otherwise communicated in a way that does not reveal any underlying data from a competitor, such as narrative industry reports, news reports, business commentaries, or generalized industry survey results.GBL § 344(1)(c).
(2)(b) 5 A coordinatorCoordinator"Coordinator" shall mean any operator of a software or data analytics service that performs a coordinating function for any landlord, including a landlord performing a coordinating function for such landlord's own benefit.GBL § 344(1)(b) shall not facilitate an agreement among landlords to not compete regarding the renting of residential premises.
(3)(a) 6 A landlord shall not share a tenant's personal data with any third party without such tenant's written consent.
(3)(b) 7 A landlord shall disclose to each tenant: (i) the categories of personal data processed by the landlord and by any processor who processes personal data on behalf of such landlord; (ii) the sources from which personal data is collected; (iii) the purposes for processing personal data; (iv) the landlord's retention period for each category of personal data that such landlord processes or which is processed on such landlord's behalf, or if that is not possible, the criteria used to determine such retention period; and (v) the identity of each third party to whom the landlord disclosed, shared, transferred, or sold personal data, and for each identified third party: (A) the categories of personal data being shared, disclosed, transferred, or sold to the third party; (B) the purposes for which personal data is being shared, disclosed, transferred, or sold to the third party; (C) the third party's retention period for each category of personal data processed by such third party or processed on such third party's behalf, or if that is not possible, the criteria used to determine such retention period; and (D) whether the third party uses the personal data for targeted advertising.
(4) In addition to any action brought by the attorney general under this article, a tenant injured by a violation of this section may bring an action to recover damages. A court may also award attorneys' fees to a prevailing plaintiff tenant.
This section prohibits landlords from using algorithmic pricing tools that incorporate nonpublic competitor data to set or change residential rent amounts, and separately prohibits coordinators from facilitating anti-competitive agreements among landlords. The provision targets the specific harm of algorithmic rent coordination — landlords sharing nonpublic occupancy, pricing, and lease data through a shared pricing algorithm or coordinator.
The section also imposes comprehensive tenant personal data disclosure requirements, requiring landlords to disclose the categories of data processed, sources, purposes, retention periods, and third-party sharing details. Sharing tenant personal data with third parties requires written consent. Enforcement includes both AG authority and a private right of action for injured tenants, with attorneys' fees available to prevailing plaintiffs.
(1)(a)–(c) "Automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a)" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making lending decisionsLending decision"Lending decision" means to screen applicants for a loan.Banking Law § 103-a(1)(c) that impact natural persons. "Automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a)" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data. (b) "Disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b)" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) shall include, but not be limited to, testing of the extent to which use of an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) shall differentiate between applicants who were approved and applicants who were not approved by the tool. (c) "Lending decisionLending decision"Lending decision" means to screen applicants for a loan.Banking Law § 103-a(1)(c)" means to screen applicants for a loan.
(2)(a)–(b) 8 2. No less than annually, each bank that uses automated decision toolsAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) to make lending decisionsLending decision"Lending decision" means to screen applicants for a loan.Banking Law § 103-a(1)(c) shall: (a) conduct a disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) to assess the actual impact of any automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) used by any bank to make lending decisionsLending decision"Lending decision" means to screen applicants for a loan.Banking Law § 103-a(1)(c) within the state; and (b) submit a summary of the most recent disparate impact analysis of such tool as well as the distribution date of the tool to which the analysis applies to the attorney general's office.
(3)(a)–(b) 9 3. (a) Any bank that uses an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) to screen applicants for a loan shall notify each such applicant of the following: (i) that an automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) will be used in connection with the assessment or evaluation of such applicant; (ii) the characteristics that such automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) will use in the assessment of such applicant; (iii) information about the type of data collected for such automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a), the source of such data, and the bank's data retention policy; and (iv) if an application for a loan is denied through use of the automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a), the reason for such denial. (b) The notice required by paragraph (a) of this subdivision shall be made no less than twenty-four hours before the use of such automated decision toolAutomated decision tool"Automated decision tool" means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making housing decisions that impact natural persons. "Automated decision tool" does not include a tool that does not automate, support, substantially assist, or replace discretionary decision-making processes and that does not materially impact natural persons, including, but not limited to, a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set, or other compilation of data.RPL § 227-g(1)(a) and shall allow such applicant to opt out of or consent to such use and/or retention of such applicant's personal information by the bank.
(3)(c) 10 If an application for a loan is denied based on personal information that is incorrect, the applicant, upon receipt of the notice required by subparagraph (iv) of paragraph (a) of this subdivision, shall have thirty days to correct such information and appeal such denial.
(4) 4. The attorney general may initiate an investigation if a preponderance of the evidence, including the summary of the most recent disparate impact analysisDisparate impact analysis"Disparate impact analysis" means an impartial evaluation conducted by an independent auditor. Such disparate impact analysis shall include, but not be limited to, testing of the extent to which use of an automated decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under this article. A disparate impact analysis shall differentiate between applicants who were selected and applicants who were not selected by the tool.RPL § 227-g(1)(b) required pursuant to paragraph (b) of subdivision two of this section, establishes a suspicion of a violation. The attorney general may also initiate in any court of competent jurisdiction any action or proceeding that may be appropriate or necessary for correction of any violation issued pursuant to this section, including mandating compliance with the provisions of this section or such other relief as may be appropriate.
This section mirrors the housing automated decision tool requirements in RPL § 227-g but applies them to banks using automated decision tools to screen loan applicants. Banks must conduct annual independent disparate impact analyses and submit summaries to the Attorney General's office — notably, the banking provision requires submission to the AG rather than public website publication. Applicant notice requirements parallel the housing section but add two significant rights: applicants may opt out of or consent to use and retention of their personal information, and applicants denied based on incorrect personal information have a 30-day right to correct the information and appeal the denial.
Enforcement is through the Attorney General, who may investigate on a preponderance of the evidence and may bring court actions. No private right of action is created for this section.
(1)–(2) 11 1. The division of housing and community renewal, in conjunction with the department of financial services and the division of human rights shall conduct a study on the impact of artificial intelligence and machine learning on housing discrimination and how such artificial intelligence and machine learning may further redlining. Such study shall be completed within one year of the effective date of this act. 2. The division of housing and community renewal shall submit a report to the governor and the legislature on its findings and recommendations for legislative action no later than ninety days after the completion of the study.
This section mandates a state study on the impact of artificial intelligence and machine learning on housing discrimination and redlining, to be conducted jointly by the Division of Housing and Community Renewal, the Department of Financial Services, and the Division of Human Rights. The study must be completed within one year of the act's effective date, and a report with findings and legislative recommendations must be submitted to the governor and legislature within 90 days of study completion. The study provision sunsets automatically upon transmission of the report.
This act shall take effect immediately; provided, however, that the provisions of section four of this act shall expire and be deemed repealed after transmission of the report of the findings and recommendations of the study to the governor and the legislature, as provided in subdivision 2 of section four of this act. Provided, further, that the commissioner of the division of housing and community renewal shall notify the legislative bill drafting commission upon the transmission of the report of the findings and recommendations of the study, as provided in subdivision 2 of section four of this act, in order that the commission may maintain an accurate and timely effective data base of the official text of the laws of the state of New York in furtherance of effectuating the provisions of section 44 of the legislative law and section 70-b of the public officers law.
The act takes effect immediately upon enactment. The study provision in section 4 expires and is deemed repealed upon transmission of the study report to the governor and legislature. The Commissioner of the Division of Housing and Community Renewal must notify the Legislative Bill Drafting Commission upon transmission.