11 dated developments from September 2026 that bear on AI compliance, each checked against the source linked. Entries are tagged by industry; the newest developments and fixed deadlines ahead are on the AI regulation news page.
Feeds (Atom): All changes · Banks and credit unions · Private equity and asset managers · Hospitals and health systems · Insurers · Employers and HR · Law firms and legal teams · Manufacturers · Importers and exporters. Data: updates.json.
Adds Business and Professions Code section 6068.1 and amends Code of Civil Procedure section 128.7, effective January 1, 2027. California attorneys may not delegate the practice of law to generative AI, must keep confidential and other nonpublic information out of AI systems unless access is restricted to people bound by confidentiality, must take reasonable steps to verify outputs, including all case and statutory citations, must disclose generative AI use to the court for all documents submitted to the court, and must consider whether to disclose it in content provided to the public.
Source: SB 574, Chapter 858, Statutes of 2026, California Legislative Information · Guide: AI compliance, layer by layer
SB 947 (Chapter 859, Statutes of 2026), a revised No Robo Bosses Act, is operative July 1, 2027. It bars an employer from relying solely on an automated decision system to discipline or terminate an employee. Where the employer relies primarily on one, a human must corroborate the output, and the employee must get a stand-alone written notice when the employer informs the employee of the decision (Labor Code sections 1522 and 1524). It does not apply under a collective bargaining agreement that explicitly waives it and meets its other conditions, or to the extent a federal statute, regulation, or binding federal contract on aircraft development or national security, military, space, or defense products requires the system or makes it reasonably necessary. From January 1, 2027, AB 1883 (Chapter 853) bars workplace surveillance tools that use AI to recognize, infer, or predict the emotional states of employees or to collect neural data; it states that it does not prohibit using a surveillance tool to ensure safety, and it exempts aircraft and defense development work where a federal statute, regulation, or binding federal contract makes the tool reasonably necessary (Labor Code sections 1581 and 1583). Also from January 1, 2027, SB 951 (Chapter 860) requires added content in Cal/WARN notices when a mass layoff, relocation, or termination is caused in whole or in substantial part by AI or other automated technology (Labor Code section 1401(d)).
Source: SB 947, Chapter 859, Statutes of 2026, California Legislative Information · Guide: AI in hiring and HR
From January 1, 2027, AB 1979 (Chapter 854) requires health facilities, clinics, and physician and group practice offices to take reasonable steps so that a licensed provider keeps independent professional judgment when care is informed by a clinical decision support system, and bars using AI to direct unlicensed personnel in, or to perform independently, a clinical function the law requires a licensed professional to perform. The AI bar does not reach use by a trainee in a supervised course of study or training program, and the section does not apply to automated documentation and communication that involves no professional judgment. SB 503 (Chapter 857) requires developers and deployers to make reasonable efforts to identify clinical decision support systems with a known or reasonably foreseeable risk of biased impacts, with mitigation, documentation, and monitoring duties.
Source: AB-1979, Chapter 854, Statutes of 2026, California Legislative Information · Guide: AI in health care
In National Fair Housing Alliance v. CFPB, No. 1:26-cv-01820, plaintiffs moved for summary judgment (ECF No. 28) against the rule removing disparate impact from the CFPB's reading of ECOA. The docket shows no motion for a preliminary injunction, so the rule remains in effect; under the July 24, 2026 scheduling order, briefing runs to January 12, 2027.
Source: Docket, National Fair Housing Alliance v. CFPB, No. 1:26-cv-01820 (D.D.C.), CourtListener RECAP · Guide: AI in financial services
On interlocutory appeal, the Third Circuit affirmed that Westlaw headnotes are original enough for copyright and that ROSS Intelligence's copying of them to train its AI legal search tool was not fair use. The court noted ROSS's tool was not generative AI and called the dispute an ordinary copyright case.
Source: Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., No. 25-2153 (3d Cir.) · Guide: AI compliance, layer by layer
MOFCOM said China and the United States had agreed to extend the October 2025 Kuala Lumpur arrangement, under which each side suspended certain measures until November 10, 2026, to January 10, 2027. The statement does not mention the November 27, 2026 end date of China's suspension of its US export ban on gallium, germanium, and antimony. As of October 21, 2026, neither MOFCOM nor BIS had published the legal text of an extension.
Source: MOFCOM Department of American and Oceanian Affairs explanation of the results of the eighth round of China-US economic and trade consultations (in Chinese), September 28, 2026 · Guide: AI compliance, layer by layer
Connected products and related services placed on the market after September 12, 2026 must be designed so that product data and related service data are, by default, easily, securely, and free of charge accessible to the user, including the metadata needed to interpret them, in a comprehensive, structured, commonly used, machine-readable format, and directly accessible where relevant and technically feasible (Article 3(1)).
Source: Regulation (EU) 2023/2854, Article 50, EUR-Lex · Guide: AI compliance, layer by layer
Proposed non-binding guidance would replace the 2023 interagency third-party guidance and ask institutions to prioritize relationships by the magnitude and likelihood of harm. Comments are due November 16, 2026.
Source: OCC Bulletin 2026-46
Manufacturers must report actively exploited vulnerabilities and severe incidents affecting the security of products with digital elements through ENISA's Single Reporting Platform, with an early warning within 24 hours and a notification within 72 hours. The duty applies to products within the Act's scope already on the market as well as new ones.
Source: Cyber Resilience Act: Reporting obligations, European Commission · Guide: AI compliance, layer by layer
SB 1119 (Chapter 190, Statutes of 2026) deletes the known-minor provisions of SB 243 from Business and Professions Code 22602, effective January 1, 2027, and adds Chapter 11.6 (commencing with Section 21810). Every operator must either determine users' ages as the statute directs or apply the chapter's child protections to all users. From July 1, 2027, before making a new or substantially modified companion chatbot available to users in California, an operator that allows child users must perform and document a comprehensive child safety risk assessment and take and document measures that reasonably mitigate the risks identified; a risk assessment conducted for another law may be used if it has a substantially similar scope and effect.
Source: SB 1119, Chapter 190, Statutes of 2026, California Legislative Information (chaptered text via bulk download) · Guide: California AI laws
Section 8 of HB 149 required the Attorney General to post the Sec. 552.102 online complaint mechanism by September 1, 2026. The Consumer AI Rights page carried a File an AI Complaint route as of October 23, 2026. A complaint received through it is what allows the Attorney General to issue a civil investigative demand under Sec. 552.103.
Source: Office of the Texas Attorney General, Consumer AI Rights · Guide: Texas AI law (HB 149)