THE AI LAW REPORTIssue 007 - Friday 18 September 2026
Published by AI Policies UK
This edition covers developments identified between 11 and 18 September 2026. It has been reconciled against Issue 006 so that earlier matters are included only where a new procedural event occurred during this reporting period. Sources were re-checked on 18 September 2026, with primary court, regulator, legislative and public-authority material used wherever obtainable.
Where a complaint, application or investigation remains unresolved, this report records the allegation or procedural step rather than treating it as a finding. Where primary documentation could not be obtained, that limitation is stated.
This week
Courts considered AI across copyright, competition, political speech and professional responsibility. The Ninth Circuit ruled on DMCA claims concerning GitHub Copilot and OpenAI Codex; a Montana federal court granted limited preliminary relief against enforcement of an election-deepfake law; a Florida appellate court ordered a lawyer to explain why sanctions should not follow filings it suspected contained AI-generated components; and a California lawyer was sanctioned after AI-assisted filings contained nonexistent authorities.
New litigation included Universal Music Group labels' copyright action against DistroKid. Existing proceedings involving xAI, OpenAI, Microsoft, Apple and ANI Media also moved forward.
Public-authority activity included two AI-related enforcement cases published by China's Cyberspace Administration, China's AI Security Governance Framework 3.0, the European Commission's proposed EU KIDS Act, a UK parliamentary report on AI and human rights, California executive action on frontier-AI oversight, U.S. federal legislative proposals and further work by the federal Judiciary on AI use.
COURTS AND LITIGATION
Ninth Circuit affirms dismissal of DMCA claims concerning GitHub Copilot and OpenAI Codex
ALR reference: ALR 2026/007/01 | Status: VERIFIED — NEW EVENT ON EXISTING MATTER
Jurisdiction: United States — federal, Ninth Circuit
Court / authority: U.S. Court of Appeals for the Ninth Circuit
Case / instrument: Doe v GitHub, Inc.
Case number / official reference: 24-7700
Area: Copyright · DMCA §1202 · Generative coding systems · Copyright-management information
Stage: Appeal decided; two contract claims remain below
Event date: 16 September 2026 | Confidence: High
What happened. The Ninth Circuit affirmed dismissal of claims brought by anonymous open-source programmers under §1202(b) of the Digital Millennium Copyright Act against GitHub, Microsoft and OpenAI-related defendants. The programmers alleged that GitHub Copilot and OpenAI Codex could reproduce code without accompanying copyright-management information.
Judge Eric Miller, writing for the panel, distinguished the creation of a new work without copyright-management information from removing or altering that information from an existing work. The court rejected the proposition that §1202 necessarily requires literal identity between the original and resulting material, but concluded that the pleaded theory did not establish the removal or alteration required by the statute.
The court did not decide a separate theory concerning alleged removal of copyright-management information from code before it was used as training material, concluding that the plaintiffs had not preserved that theory. Two breach-of-contract claims remain in the district court. (CA9 Courts CDN)
The underlying question. Whether outputs generated by Copilot or Codex without copyright-management information support liability under DMCA §1202(b).
What this development does not establish. The decision does not determine whether AI training on the plaintiffs' code infringed copyright. Nor does it determine whether Copilot outputs constitute ordinary copyright infringement.
What happens next. The surviving contract claims continue in the Northern District of California.
Verification. Verified against the Ninth Circuit's published opinion.
Source: Ninth Circuit opinion — Doe v GitHub, No. 24-7700
Florida appellate court issues show-cause order over suspected AI-assisted filings
ALR reference: ALR 2026/007/02 | Status: VERIFIED — JUDICIAL ORDER / POSSIBLE PROFESSIONAL SANCTIONS
Jurisdiction: United States — Florida
Court / authority: Florida Fourth District Court of Appeal
Case / instrument: Lorie Jean Lisandrillo v Renee Palozzi
Case number: 4D2026-2262
Area: Appellate procedure · Professional responsibility · AI-assisted legal drafting
Stage: Writ petition denied; sanctions question pending
Event date: 16 September 2026 | Confidence: High
What happened. The Fourth District Court of Appeal denied an extraordinary writ petition in a family-law proceeding and separately ordered petitioner's counsel, Jaclyn R. Soroka, to show cause why sanctions should not be imposed for frivolous filings.
The court said it suspected the filings contained AI-generated components. Importantly, this was not a fabricated-citation case: the court expressly said the cited cases existed. Its concern was the quality and structure of the legal reasoning, including what it described as convoluted, false and frivolous arguments and apparent absence of meaningful independent professional judgment.
The opinion linked the conduct to counsel's professional obligations, including the Florida rule requiring independent professional judgment.
Counsel was ordered to respond within ten days — expressly without using AI — and explain why sanctions, potentially including referral to the Florida Bar, should not be imposed. (Root)
The underlying question. Whether counsel's filings constituted frivolous or abusive appellate practice warranting sanctions under Florida appellate rules.
What this development does not establish. The court has not yet imposed sanctions. It said it suspected AI involvement; the opinion does not establish which AI system was used or conclusively determine how the documents were produced.
What happens next. Counsel must respond to the show-cause order within ten days of the opinion.
Verification. Verified against the Florida Fourth District's published opinion.
Source: Florida Fourth District opinion
California lawyer sanctioned $999.99 after AI-assisted filings contained nonexistent authorities
ALR reference: ALR 2026/007/03 | Status: VERIFIED — NEW EVENT ON EXISTING MATTER
Jurisdiction: United States — California
Court / authority: Los Angeles County Superior Court; Judge Elizabeth L. Bradley
Case / instrument: Fa'alagilagi Meni-Siliga v A's Contractor Inc. et al.
Case number: 24CMCV01105
Area: Civil procedure · Professional responsibility · AI-assisted legal research
Stage: Monetary sanction imposed; underlying insurance litigation continues
Event date: 11 September 2026 | Confidence: High
What happened. Attorney Jacquelene A. Robinson, outside counsel representing State Farm General Insurance Company, was ordered to pay $999.99 following AI-assisted filings containing seven nonexistent case citations.
Robinson had previously acknowledged using an AI legal-research system called Irys while preparing pre-trial motions and failing independently to verify some of the resulting authorities. The citation problems were identified by opposing counsel.
Judge Bradley imposed the monetary sanction following the court's order-to-show-cause process. (MLex)
The underlying question. Whether counsel's submission of filings containing nonexistent authorities warranted monetary sanctions.
What this development does not establish. The sanction concerns counsel's court submissions. It does not determine State Farm's liability in the underlying property-insurance dispute.
What happens next. The underlying litigation continues, with trial previously listed for 5 October 2026.
Verification. Counsel's earlier declaration establishes the AI-assisted citation failures; the subsequent monetary sanction is independently corroborated. The complete sanctions order was not separately retrieved for this edition.
xAI seeks Eighth Circuit injunction against Minnesota AI nudification law
ALR reference: ALR 2026/007/04 | Status: VERIFIED — NEW EVENT ON EXISTING MATTER
Jurisdiction: United States — federal, Minnesota / Eighth Circuit
Court / authority: U.S. Court of Appeals for the Eighth Circuit
Case: X.AI LLC v Keith M. Ellison
Appellate number: 26-2806
Area: First Amendment · Generative imagery · AI regulation
Stage: Appeal active; injunction-pending-appeal motion filed
Event date: 11 September 2026 | Confidence: High
What happened. xAI filed a motion in the Eighth Circuit seeking an injunction pending appeal against enforcement of Minnesota Stat. §325E.91, which regulates services capable of generating specified sexualised or “nudified” imagery.
The district court had refused a preliminary injunction on 4 September. Judge Donovan Frank did not decide the ultimate First Amendment question; he concluded at that stage that xAI had not demonstrated the necessary irreparable harm and that the balance of harms favoured the state.
The Eighth Circuit docket records xAI's injunction-pending-appeal motion on 11 September. (Justia Dockets & Filings)
The underlying question. Whether xAI should receive interim protection from enforcement while its constitutional challenge proceeds.
What this development does not establish. Neither the district court's refusal of preliminary relief nor the appellate application establishes whether the Minnesota law is constitutional.
What happens next. The appellate motion remains to be determined. The Eighth Circuit docket lists the appellant's principal brief as due on 29 October 2026. (Justia Dockets & Filings)
Apple leaves X/SpaceXAI antitrust litigation; OpenAI later denied access to confidential resolution agreement
ALR reference: ALR 2026/007/05 | Status: VERIFIED — MULTIPLE NEW EVENTS ON EXISTING MATTER
Jurisdiction: United States — federal, Texas
Court: U.S. District Court for the Northern District of Texas; Judge Mark T. Pittman
Matter: X Corp. and SpaceXAI LLC v Apple Inc., OpenAI Foundation et al.
Case number: 4:25-cv-00914-P
Area: Antitrust · Generative-AI chatbot competition
Stage: Apple claims concluded; litigation against OpenAI continues
Event dates: 14–18 September 2026 | Confidence: High
What happened. X Corp. and SpaceXAI resolved their claims against Apple and obtained dismissal of those claims with prejudice. The dismissal did not extend to the OpenAI defendants.
The underlying action alleges anticompetitive conduct involving smartphone and generative-AI chatbot markets, including allegations concerning ChatGPT's integration into Apple Intelligence and treatment of competing chatbot services.
A further dispute then developed over the confidential agreement resolving the Apple claims. OpenAI sought access to that agreement for use in defending the surviving antitrust action. On 18 September, Judge Pittman denied OpenAI's request to inspect it. (Reuters)
The underlying question. The surviving case concerns the plaintiffs' antitrust allegations against OpenAI. The narrower 18 September question concerned whether OpenAI should receive access to the confidential Apple resolution.
What this development does not establish. Apple's dismissal is not an adjudicated finding that Apple won the antitrust allegations on their merits. The terms of the Apple resolution remain confidential. The 18 September order does not determine the substantive antitrust claims against OpenAI.
What happens next. Litigation against the OpenAI defendants continues.
Universal Music labels file copyright action against DistroKid
ALR reference: ALR 2026/007/06 | Status: VERIFIED — NEW MATTER
Jurisdiction: United States — federal, Delaware
Court: U.S. District Court for the District of Delaware
Case: UMG Recordings, Inc. et al. v DistroKid, LLC et al.
Case number: 1:26-cv-01156
Area: Copyright · AI-generated music · Distribution
Stage: Complaint filed and defendants served
Event date: 15 September 2026 | Confidence: High as to filing; allegations unresolved
What happened. UMG Recordings, Capitol Records and Capitol CMG sued DistroKid and related entities for copyright infringement and related claims.
The complaint includes allegations concerning distribution of mass-generated AI music as well as alleged distribution of conventional infringing recordings. Reporting based on the complaint also identifies alleged use of the Suno music-generation system.
The federal docket confirms that the complaint was filed on 15 September and that the defendants were served that day. Their responses were listed as due 6 October 2026. (Justia Dockets & Filings)
The underlying question. Whether the defendants' alleged distribution and associated conduct infringed the plaintiffs' copyrights and supports the other causes of action pleaded.
What this development does not establish. Filing the complaint does not establish copyright infringement or any other liability. The action is not, on the material reviewed, a claim that every distribution of AI-generated music is inherently unlawful.
What happens next. Responsive pleadings are currently due 6 October.
Verification. Federal docket and complaint filing verified; substantive allegations remain allegations.
Montana federal court grants limited preliminary injunction in election-deepfake challenge
ALR reference: ALR 2026/007/07 | Status: VERIFIED — PRELIMINARY INJUNCTION
Jurisdiction: United States — federal, Montana
Court: U.S. District Court for the District of Montana; Judge Susan Watters
Area: First Amendment · Elections · AI-generated political media
Stage: Preliminary relief granted to named plaintiffs
Event date: 17 September 2026 | Confidence: High
What happened. Judge Susan Watters granted preliminary relief to plaintiffs challenging Montana's Senate Bill 25, which regulates specified AI-generated or manipulated election material.
The dispute followed complaints concerning political mailers containing AI-generated imagery. In granting preliminary relief, the court concluded that the plaintiffs were likely to succeed on their First Amendment challenge, including their argument that the statute discriminated according to viewpoint by treating certain AI-generated material differently depending on whether it supported or harmed a candidate.
The court did not impose a universal injunction. Relief was limited to the plaintiffs, with the court considering the U.S. Supreme Court's restrictions on universal injunctions. (Reuters)
The underlying question. Whether Montana's restrictions on AI-generated election material are consistent with the First Amendment.
What this development does not establish. This is preliminary relief, not a final judgment invalidating the statute statewide.
What happens next. The constitutional litigation continues.
Delhi High Court issues notice to OpenAI in ANI copyright appeal
ALR reference: ALR 2026/007/08 | Status: PARTIALLY VERIFIED — NEW EVENT ON EXISTING MATTER
Jurisdiction: India
Court: Delhi High Court, Division Bench
Matter: ANI Media Pvt Ltd v OpenAI OpCo LLC
Underlying case: CS(COMM) 1028/2024
Area: Copyright · Generative-AI training · Interim injunctions
Stage: Appeal proceeding; OpenAI response sought
Event date: 15 September 2026 | Confidence: Medium-high
What happened. Contemporary court reporting records that a Division Bench issued notice to OpenAI in ANI's appeal against the July refusal of interim injunctive relief.
ANI sought immediate protection connected with OpenAI's alleged scraping and use of its journalism. The Division Bench declined to impose the requested restraint without first hearing OpenAI and sought OpenAI's response.
The underlying question. Whether ANI should receive interim copyright protection pending resolution of its claims concerning alleged use of its material in connection with generative AI.
What this development does not establish. Issuing notice does not indicate acceptance of ANI's copyright arguments. Refusal of ex parte relief is likewise not a determination that OpenAI's conduct is lawful.
What happens next. The matter is reported as listed for 5 December 2026.
Verification. Multiple contemporaneous reports support the procedural event. The Division Bench's primary order was not obtained before editorial cut-off.
OpenAI copyright MDL: Magistrate Judge Wang discloses Microsoft holding and remains assigned
ALR reference: ALR 2026/007/09 | Status: VERIFIED — BACKFILLED EVENT ON EXISTING MATTER
Jurisdiction: United States — federal, New York
Court: U.S. District Court for the Southern District of New York
Matter: In re OpenAI, Inc. Copyright Infringement Litigation
Case number: 1:25-md-03143
Area: Judicial recusal · Copyright litigation · Generative AI
Stage: Consolidated pre-trial proceedings continue
Event date: 11 September 2026; publicly reported 14 September | Confidence: High
What happened. U.S. Magistrate Judge Ona T. Wang disclosed that she had owned Microsoft stock while overseeing pre-trial issues in the consolidated copyright litigation involving OpenAI and Microsoft.
Wang said she had divested the interest and that it had not affected her decisions. She concluded that she could remain assigned and referred to the public interest in avoiding reassignment of the extensive consolidated proceedings. (Reuters)
The underlying question. Whether the financial interest required Wang's recusal notwithstanding divestment.
What this development does not establish. The disclosure does not determine any copyright, fair-use, training-data or damages question in the underlying litigation.
What happens next. Consolidated pre-trial proceedings continue.
COURT PROCEDURE AND JUDICIAL ADMINISTRATION
Tenth Circuit proposes human-review certification for AI-assisted filings
ALR reference: ALR 2026/007/10 | Status: VERIFIED — PROPOSED COURT RULE
Jurisdiction: United States — federal
Court / authority: U.S. Court of Appeals for the Tenth Circuit
Instrument: Proposed revisions to 2027 Tenth Circuit Rules
Area: Court procedure · Generative AI · Professional responsibility
Stage: Public-comment process / proposed rule
Event date: 18 September 2026 | Confidence: High
What happened. The Tenth Circuit opened its process for revisions to its 2027 local rules. The proposed changes include certification requirements concerning filings prepared with generative-AI assistance, requiring human review for accuracy.
The court states that its revised rules are intended to take effect on 1 January 2027. (Reuters)
The underlying question. What procedural assurance should accompany AI-assisted documents filed in the Tenth Circuit.
What this development does not establish. The proposal is not yet an operative AI filing requirement.
What happens next. Public comment and finalisation of the 2027 rules.
Source: Tenth Circuit rule-revision notice
U.S. Judicial Conference receives update on federal Judiciary's AI work
ALR reference: ALR 2026/007/11 | Status: VERIFIED — JUDICIAL ADMINISTRATION
Jurisdiction: United States — federal judiciary
Authority: Judicial Conference of the United States / Administrative Office of the U.S. Courts
Area: Judicial administration · AI use · Court policy
Stage: AI task-force work continuing
Event date: 17 September 2026 | Confidence: High
What happened. The Judicial Conference received an update on the federal Judiciary's work addressing AI.
Administrative Office Director Judge Robert J. Conrad Jr. reported that the Judiciary's AI task force has identified more than 60 distinct issues and formed seven subject-matter groups. Interim guidance has cautioned courts against delegating core judicial functions, including decision-making and case adjudication, to AI and reminded Judiciary users that they remain accountable for AI-assisted work. (United States Courts)
What this development does not establish. The update is not a new Federal Rule of Procedure or statute, and not every issue identified by the task force has resulted in final guidance.
What happens next. Additional guidance is expected as the task force completes work on individual issues.
Source: U.S. Courts — Judicial Conference update
REGULATORS AND ENFORCEMENT
China orders AI mini-program offline for generated-content labelling and security-assessment failures
ALR reference: ALR 2026/007/12 | Status: VERIFIED — REGULATORY ENFORCEMENT
Jurisdiction: China
Authority: Cyberspace Administration system / local cyberspace authority
Area: Generative AI · Synthetic-content labelling · Security assessment
Stage: Enforcement action announced
Event date: Published 15 September 2026 | Confidence: High
What happened. China's Cyberspace Administration published an enforcement case involving a Sichuan technology company whose WeChat mini-program provided AI text dialogue and image generation.
According to CAC, the service failed to add required visible labels to AI-generated or synthetic material and failed to place required information in file metadata. CAC also said required security-assessment obligations had not been met.
The relevant cyberspace authority ordered the mini-program taken offline. CAC identified the Deep Synthesis Provisions, the Measures for Labelling AI-Generated and Synthetic Content and security-assessment rules among the requirements breached. (China Anti-Corruption Agency)
The underlying question. Compliance with China's requirements for identification of AI-generated content and security assessment of covered AI services.
What this development does not establish. CAC's published example does not identify the company by full name and should not be attributed to an unverified provider.
Source: CAC enforcement cases, 15 September
China warns API intermediary after generative-AI security-assessment failure
ALR reference: ALR 2026/007/13 | Status: VERIFIED — REGULATORY ENFORCEMENT
Jurisdiction: China
Authority: Cyberspace Administration system
Area: Generative AI · APIs · Security assessment
Event date: Published 15 September 2026 | Confidence: High
What happened. A second AI-specific example in CAC's enforcement release concerns a Jiangsu technology company operating two websites that acted as API intermediaries for multiple large-model products.
CAC said the company offered dialogue and question-answering services without completing required security assessments. The local authority ordered rectification, dealt with responsible personnel and imposed a warning. (China Anti-Corruption Agency)
What this development does not establish. The announcement does not identify the large-model providers whose APIs were accessed and does not establish misconduct by those underlying providers.
Source: CAC enforcement cases, 15 September
Manhattan DA seizes domains of 12 websites in AI-generated intimate-image investigation
ALR reference: ALR 2026/007/14 | Status: VERIFIED — INVESTIGATIVE ENFORCEMENT
Jurisdiction: United States — New York
Authority: Manhattan District Attorney's Office / New York State court
Area: Criminal investigation · Non-consensual intimate imagery · AI deepfakes
Stage: Court-authorised seizures; investigation ongoing
Event date: 14 September 2026 | Confidence: High
What happened. The Manhattan District Attorney announced the seizure, pursuant to court order, of 12 domain names associated with websites under investigation for allegedly publishing, distributing and selling AI-generated non-consensual intimate imagery.
Prosecutors allege that AI image and video tools were used to transform existing images of approximately 1,200 real people into synthetic intimate material. The investigation remains ongoing. (Manhattan District Attorney's Office)
The underlying question. Whether the investigated creation, distribution and sale of the material amounts to criminal conduct and supports further enforcement.
What this development does not establish. Domain seizure is an investigative measure, not a conviction. The District Attorney's account of the underlying conduct remains an allegation.
What happens next. The Cyber Crime Bureau's investigation continues.
Source: Manhattan District Attorney announcement
LEGISLATION AND PROPOSALS
European Commission proposes EU KIDS Act covering AI companions and chatbots
ALR reference: ALR 2026/007/15 | Status: VERIFIED — LEGISLATIVE PROPOSAL
Jurisdiction: European Union
Authority: European Commission
Instrument: EU KIDS Act — EU Keeping Internet Digital Spaces Accountable and Trustworthy
Official reference: COM(2026) 681 final; 2026/0286(COD)
Area: Child online safety · AI companions · Chatbots · Digital services
Stage: Commission proposal; ordinary legislative procedure
Event date: 17 September 2026 | Confidence: High
What happened. The European Commission adopted a proposal for a regulation covering online services used by minors, expressly including AI companions and general conversational chatbots.
The proposal establishes age-based restrictions for social media and safety-by-design requirements for services used by minors. The Commission states that AI chatbots and companions would be off by default for minors and subject to requirements addressing conduct capable of creating emotional dependency. (European Commission)
The underlying question. What EU-wide obligations should govern children's access to and interaction with specified online services and AI systems.
What this development does not establish. The KIDS Act is not yet EU law. Its final text may change through the legislative process.
What happens next. The European Parliament and Council will consider and negotiate the proposal. (Digital Strategy)
Sources: European Commission KIDS Act proposal | EUR-Lex COM(2026) 681 final
UK Joint Committee on Human Rights calls for AI legislation
ALR reference: ALR 2026/007/16 | Status: VERIFIED — PARLIAMENTARY REPORT
Jurisdiction: United Kingdom
Authority: Joint Committee on Human Rights
Instrument: Human Rights and the Regulation of AI
Official reference: Fourth Report of Session 2026–27; HC 160 / HL Paper 56
Area: Human rights · AI regulation · Automated decision-making
Stage: Committee recommendations to Government
Event date: 14 September 2026 | Confidence: High
What happened. Parliament's Joint Committee on Human Rights published its report examining AI and the UK's human-rights framework. The report recommends legislation and sets out proposed safeguards concerning areas including high-risk AI, transparency, oversight, remedies and specified uses of AI.
Parliament expressly describes the document as recommendations to government and states that the Government has two months to respond. (UK Parliament Committees)
The underlying question. Whether the UK's existing legal and regulatory framework provides sufficient protection for human rights where AI systems are developed and deployed.
What this development does not establish. The Committee's recommendations are not legislation and do not themselves create a regulator or new statutory obligations.
What happens next. Government response expected under the parliamentary process.
Source: UK Parliament — Human Rights and the Regulation of AI
U.S. Senate unanimous-consent attempt on AI Emergency Button Act is blocked
ALR reference: ALR 2026/007/17 | Status: VERIFIED — LEGISLATIVE PROCEDURAL EVENT
Jurisdiction: United States — federal
Authority: U.S. Senate
Instrument: AI Emergency Button Act
Area: Frontier AI · Emergency shutdown capability
Stage: Unanimous-consent attempt unsuccessful
Event date: 16 September 2026 | Confidence: High
What happened. Senator John Kennedy attempted to pass the AI Emergency Button Act by unanimous consent. Senator Rand Paul objected, preventing passage through that procedure.
Kennedy's proposal would require covered AI developers to provide an emergency shutdown mechanism. (John Kennedy)
What this development does not establish. The Senate did not vote on final passage and did not reject every possible version of an AI shutdown requirement. The specific event was an objection to passage by unanimous consent.
Source: U.S. Senator John Kennedy — 16 September procedural announcement
Covered AI Prohibition Act introduced in U.S. House
ALR reference: ALR 2026/007/18 | Status: VERIFIED — LEGISLATIVE PROPOSAL
Jurisdiction: United States — federal
Authority: U.S. House of Representatives
Instrument: H.R.10484 — Covered AI Prohibition Act
Area: Defence systems · AI procurement / exclusion
Stage: Introduced and referred to House Armed Services Committee
Event date: 17 September 2026 | Confidence: High
What happened. Representative Don Bacon introduced H.R.10484. Its official title states that it would require the Secretary of Defense to issue Department of Defense-wide guidance for identifying specified AI products and excluding and removing those products from Defense systems and devices.
The bill was referred to the House Armed Services Committee. (GovInfo)
What this development does not establish. Introduction does not enact the proposal or establish that any AI product has been excluded under the proposed mechanism.
What happens next. Committee consideration.
Source: GovInfo — H.R.10484
AIM to Cure Kids with Cancer Act introduced
ALR reference: ALR 2026/007/19 | Status: VERIFIED — LEGISLATIVE PROPOSAL
Jurisdiction: United States — federal
Authority: U.S. House of Representatives
Instrument: Ailani Myers Accelerating Innovation in Medicine (AIM) to Cure Kids with Cancer Act
Area: Healthcare · Paediatric cancer · AI research and data
Stage: Bill introduced
Event date: 14 September 2026 | Confidence: High
What happened. Representatives Michael McCaul, Ami Bera and Mike Kelly announced legislation intended to accelerate the use of AI in paediatric cancer research, treatment development and data storage. The sponsors describe it as building on an earlier executive initiative. (Congressman Michael McCaul)
What this development does not establish. The proposal has not been enacted.
Verification. Introduction verified through the sponsor's official congressional announcement. The complete bill number was not independently confirmed in the material reviewed.
Source: Congressman Michael McCaul — bill announcement
GOVERNMENT AND PUBLIC AUTHORITIES
China releases Artificial Intelligence Security Governance Framework 3.0
ALR reference: ALR 2026/007/20 | Status: VERIFIED — GOVERNANCE FRAMEWORK
Jurisdiction: China — national
Authority: National Cybersecurity Standardisation Technical Committee (TC260), under CAC guidance
Instrument: Artificial Intelligence Security Governance Framework 3.0
Area: AI security · Risk classification · Technical countermeasures
Stage: Framework released
Event date: 14 September 2026 | Confidence: High
What happened. TC260 released Artificial Intelligence Security Governance Framework 3.0 at China's 2026 National Cybersecurity Awareness Week.
The updated framework retains a structure based on risk classification, technical countermeasures and governance measures while updating risk categories and responses to reflect newer AI developments. (Fudan University)
What this development does not establish. The framework should not be described as a court judgment or a new comprehensive AI statute. Its legal and administrative status must be distinguished from binding legislation.
China publishes new generative-AI service filing figures
ALR reference: ALR 2026/007/21 | Status: VERIFIED — REGULATORY ADMINISTRATION
Jurisdiction: China
Authority: Cyberspace Administration of China
Area: Generative AI · Registration / filing
Stage: Filing information published
Event date: 14 September 2026 | Confidence: High
What happened. CAC published information covering 124 additional generative-AI services filed during July and August, including seven on-device services. A further 133 applications or functions using registered models through APIs or other means completed local registration.
The published figures brought the totals reported by CAC at 31 August to 1,112 generative-AI services filed and 731 applications/functions registered. (Digital Policy Alert)
What this development does not establish. Filing or registration does not amount to a judicial finding concerning a service's legality or quality.
China publishes “AI + Software” Special Action Implementation Plan
ALR reference: ALR 2026/007/22 | Status: VERIFIED — POLICY INSTRUMENT / BACKFILL
Jurisdiction: China
Authority: Ministry of Industry and Information Technology
Instrument: “Artificial Intelligence + Software” Special Action Implementation Plan
Official reference: MIIT [2026] No. 209
Area: Software policy · AI agents · AI-enabled services · Industrial development
Stage: Plan issued and publicly explained
Publication development: 11 September 2026 | Confidence: High
What happened. MIIT published and explained its “AI + Software” Special Action Implementation Plan, setting out measures intended to deepen integration between AI and China's software and information-services sector.
MIIT's 11 September materials describe development of intelligent programming, agent software, AI-enabled services, standards, datasets, computing support, open-source projects and associated safety work. (MIIT)
What this development does not establish. The 11 September date reflects the public publication/explanatory development captured in this edition and should not be used to imply that every underlying policy decision was first made that day.
Source: MIIT explanation of the AI + Software plan
California executive order accelerates AI oversight work and directs development of further frontier-AI proposals
ALR reference: ALR 2026/007/23 | Status: VERIFIED — EXECUTIVE ORDER
Jurisdiction: United States — California
Authority: Governor of California / Government Operations Agency
Area: Frontier-AI safety · Independent verification · Emergency controls
Stage: Executive order issued
Event date: 18 September 2026 | Confidence: High
What happened. Governor Gavin Newsom issued an executive order directing state agencies to accelerate implementation of California's recently enacted independent AI-verification and auditor framework and convene experts to recommend possible additional measures.
The order directs consideration of proposals including independent verification of frontier-AI safety material and development of an emergency shutoff or “kill switch” for frontier models. (Governor of California)
What this development does not establish. The executive order does not itself impose a mandatory kill-switch requirement on frontier-AI companies. It directs work on recommendations for potential changes to state law and implementation of existing measures.
What happens next. The Governor's announcement states that experts are to provide a guide within two months.
Source: California Governor — 18 September executive order
UK and U.S. laboratories sign SUNRISE–STELLAR-AI declaration
ALR reference: ALR 2026/007/24 | Status: VERIFIED — PUBLIC-AUTHORITY COOPERATION
Jurisdiction: United Kingdom / United States
Authorities: UK Atomic Energy Authority; Princeton Plasma Physics Laboratory
Instrument: SUNRISE–STELLAR-AI Federation Joint Declaration of Intent
Area: AI · Supercomputing · Fusion research
Stage: Declaration signed; technical federation being explored
Event date: 14 September 2026 | Confidence: High
What happened. UKAEA and PPPL signed a Joint Declaration of Intent to explore linking UKAEA's SUNRISEmission-focused AI supercomputer with PPPL's STELLAR-AI platform.
The proposed federation concerns AI, high-performance computing and fusion research, with potential work including digital twins and movement of computing workloads between platforms. (GOV.UK)
What this development does not establish. The declaration does not mean that the systems are already technically integrated.
Source: UKAEA — SUNRISE–STELLAR-AI announcement
Global Fusion Policy Summit publishes joint statement addressing AI among international cooperation priorities
ALR reference: ALR 2026/007/25 | Status: VERIFIED — INTERNATIONAL PUBLIC-AUTHORITY STATEMENT
Jurisdiction: International / UK-hosted
Authorities: UK Department for Energy Security and Net Zero; U.S. Department of Energy and participating governments
Instrument: Global Fusion Policy Summit joint statement
Area: Fusion regulation · International cooperation · AI
Event date: 14 September 2026 | Confidence: High
What happened. The Global Fusion Policy Summit produced a joint statement setting out shared priorities for international fusion cooperation. The UK Government records AI among the subjects considered alongside regulation, public-private collaboration, insurance, regulatory harmonisation and standards. (GOV.UK)
What this development does not establish. The statement is not itself an AI regulatory instrument and does not create generally binding AI obligations.
Canada and Germany announce planned investment in LawZero
ALR reference: ALR 2026/007/26 | Status: VERIFIED — GOVERNMENT FUNDING ACTION
Jurisdiction: Canada / Germany
Authorities: Government of Canada; Federal Government of Germany
Area: AI research · AI safety · Sovereign technology
Stage: Planned public investment announced
Event date: 16 September 2026 | Confidence: High
What happened. Canada and Germany announced plans to invest CAD150 million and €100 million respectively in LawZero for work on advanced AI designed around transparency, trust and safety.
The Canadian announcement states that the German funding remains subject to European Commission notification. (Canada)
What this development does not establish. The announcement is public research/technology funding, not AI regulation and not approval of a particular model for deployment.
Source: Government of Canada announcement
Canada and Germany hold first Sovereign Technology Alliance co-chairs meeting
ALR reference: ALR 2026/007/27 | Status: VERIFIED — INTERGOVERNMENTAL COOPERATION
Jurisdiction: Canada / Germany
Authorities: Canadian Ministry of Artificial Intelligence and Digital Innovation; German Federal Ministry for Digital Transformation and Government Modernisation
Area: Sovereign AI · Digital infrastructure · International cooperation
Event date: 16 September; statement published 17 September 2026 | Confidence: High
What happened. Canadian and German ministers held the first co-chairs meeting of the Sovereign Technology Alliance, taking stock of prospective membership and work towards a future launch with founding members.
The governments describe the Alliance as a platform through which Canada, Germany and trusted partners intend to cooperate on sovereign AI capabilities and other technologies. (Canada)
What this development does not establish. The meeting did not create binding AI regulation.
Hong Kong establishes work on minors' use of AI and social media
ALR reference: ALR 2026/007/28 | Status: VERIFIED — GOVERNMENT REVIEW
Jurisdiction: Hong Kong
Authority: Hong Kong Government / Department of Justice-led work
Area: Child safety · AI interaction · Social media
Stage: Policy review; possible measures under examination
Event date: 18 September 2026 | Confidence: High
What happened. Hong Kong confirmed work to examine risks associated with minors' use of social media and AI and to consider possible responses.
Secretary for Justice Paul Lam said overseas regulatory approaches would be studied and emphasised that legislation is only one possible option, alongside measures including education. He also said the eventual approach would need to reflect Hong Kong's circumstances. (Hong Kong Government News)
What this development does not establish. Hong Kong has not announced enacted AI legislation through this review, and no final regulatory model has yet been selected.
Source: Hong Kong Government — AI safety for minors
DEVELOPMENTS TO WATCH
The Florida Fourth District Court of Appeal has given counsel in Lisandrillo v Palozzi ten days to respond to its sanctions show-cause order. (Root)
In UMG Recordings v DistroKid, the federal docket currently records 6 October 2026 as the defendants' response deadline. (Justia Dockets & Filings)
The underlying Meni-Siliga insurance litigation is currently recorded for trial on 5 October 2026.
In X.AI LLC v Ellison, the Eighth Circuit docket records the appellant's brief as due 29 October 2026; the injunction-pending-appeal request remains a separate matter requiring monitoring. (Justia Dockets & Filings)
ANI Media v OpenAI is reported as returning to the Delhi High Court on 5 December 2026. Because the primary appellate order was not obtained for this edition, that date should remain qualified as reported.
The EU KIDS Act now proceeds to the European Parliament and Council. It remains a Commission proposal rather than operative EU law. (EUR-Lex)
The UK Government has two months to respond to the Joint Committee on Human Rights report on AI regulation. (UK Parliament)
California's 18 September executive order calls for expert recommendations within two months on possible further frontier-AI safety measures. (Governor of California)
The Manhattan District Attorney's investigation into AI-generated non-consensual intimate imagery remains ongoing. (Manhattan District Attorney's Office)
PENDING VERIFICATION — NOT TREATED AS ESTABLISHED DEVELOPMENTS
Several matters identified during discovery have deliberately not been promoted into the substantive report.
The reported TRE-CE Brazilian electoral-advertising appeal remains pending because the official judgment, complete case number, parties, judgment date and fine have not been obtained.
Reported prospective TRAI amendments concerning AI-generated or automated commercial calls in India remain outside the verified report because no final TRAI notification was obtained.
The reported UN High Commissioner for Human Rights communication concerning frontier AI has not been included as an established new instrument because the underlying 14 September communication was not obtained.
The reported 15 September class-certification motion in Mobley v Workday remains held pending retrieval of the primary filing.
These matters remain candidates for backfill rather than being silently converted into established events.
The AI Law Report reports legal, regulatory and public-authority developments for information and educational purposes. It does not provide legal advice or predict legal outcomes.
Next edition: Friday 25 September 2026.