THE AI LAW REPORT Issue 009 - Friday 2 October 2026
Global developments from the courts, regulators and public authorities
Issue 009 | Friday 2nd October 2026
Published by AI Policies UK
This Week
The Third Circuit’s ruling on ROSS and the dismissal of Chegg’s and Penske Media’s antitrust claims lead this edition. Courts in Japan and India addressed the use of personal identity, while litigation in the United States and European Union concerns synthetic images, AI safety and access for competing services.
California enacted rules for lawyers using generative AI and a broader package of AI legislation. Connecticut’s AI-related layoff disclosure requirement took effect, South Australia’s AI Royal Commission commenced, and Brazil entered its election-period restriction on new synthetic content. The edition also covers California’s investigative subpoena to OpenAI, FTC scrutiny, Australian parliamentary scrutiny, the Bank of England’s risk assessment and the White House’s voluntary frontier-model commitments.
Courts & Litigation
Third Circuit rules for Thomson Reuters in ROSS copyright appeal
Jurisdiction: United States — federal
Court / authority: US Court of Appeals for the Third Circuit
Case / instrument: Thomson Reuters Enterprise Centre GmbH and West Publishing Corp v ROSS Intelligence Inc; 25-2153
Area: Copyright · AI training · Legal research
Stage and date: Interlocutory appeal decided · 29 September 2026; opinion publicly available 30 September
What happened. The court affirmed the ruling concerning ROSS’s use of 2,243 Westlaw headnotes to develop a competing legal search tool. It held that the editorial headnotes were sufficiently original for copyright protection and rejected ROSS’s fair-use defence. Its reasoning addressed commercial purpose, limited transformation and competition with Westlaw.
The underlying question. Whether the Westlaw headnotes were protected by copyright and whether ROSS’s use of them to develop a competing legal search tool qualified as fair use.
What this establishes. The opinion distinguishes ROSS’s search platform from models capable of generating original expression. It does not determine the legality of every generative-AI training workflow.
What happens next. On 2 October, LawSites reported that ROSS’s counsel intended to seek Supreme Court review. This is an announced intention; it does not establish that a petition has been filed or that review has been granted.
Source: Third Circuit opinion, No 25-2153
Source: LawSites, 2 October: statement from ROSS’s counsel
Court dismisses Chegg and Penske Media claims over Google AI search
Jurisdiction: United States — federal
Court / authority: US District Court for the District of Columbia; Judge Amit P Mehta
Case / instrument: Chegg Inc v Google LLC; Penske Media Corporation and others v Google LLC; 25-cv-00543 (APM); 25-cv-03192 (APM)
Area: Competition · Generative search · Unjust enrichment
Stage and date: Motions to dismiss granted · 30 September 2026
What happened. Both amended complaints were dismissed. The court found that the publishers had not plausibly alleged the reciprocal bargain underlying their reciprocal-dealing claims. Penske’s tying claim also failed to plead separate products adequately. Other antitrust claims failed, and the court declined supplemental jurisdiction over the state-law unjust-enrichment claims.
The underlying question. Whether the publishers’ amended complaints adequately pleaded antitrust claims concerning Google’s search practices and use of their content in AI-generated answers.
What this establishes. This is not a general ruling that AI Overviews, training on publisher content or every Google practice is lawful. The unjust-enrichment claims were not adjudicated on their merits. The memorandum opinion states that a final, appealable order accompanies it.
Source: Memorandum opinion, Penske ECF 32 / Chegg litigation
Tokyo court discusses voice protection but dismisses removal claim
Jurisdiction: Japan
Court / authority: Tokyo District Court, Civil Division 46
Case / instrument: Kenjiro Tsuda v TikTok Pte Ltd; Reiwa 7 (Wa) No 70635
Area: Publicity rights · Synthetic voice · Removal remedies
Stage and date: First-instance judgment; claims dismissed · 30 September 2026
What happened. Tsuda sought removal of 188 posts containing narration he said resembled his voice. The court stated that unauthorised use of a voice can infringe publicity rights where it has customer-attracting value and the use exclusively exploits that value. However, it found the account data had already been deleted and dismissed the removal claims.
The underlying question. Whether unauthorised use of a recognisable voice can engage publicity rights, and whether removal relief remained available after deletion of the relevant account data.
What this establishes. The court dismissed the removal claims because the relevant data had already been deleted. It did not determine the remaining infringement issues or order TikTok to remove the posts.
Source: Tokyo District Court judgment, official PDF
Bombay High Court grants interim protection to Samantha Ruth Prabhu
Jurisdiction: India
Court / authority: Bombay High Court; Justice Madhav Jamdar
Case / instrument: Samantha Ruth Prabhu v John Doe (Ashok Kumar) & Ors; IA (L) 29905 of 2026 in Commercial IP Suit 346 of 2026
Area: Personality rights · Publicity rights · AI deepfakes
Stage and date: Ad-interim relief granted · 1 October 2026
What happened. Justice Madhav J Jamdar granted interim protection against misuse of Prabhu’s identity in deepfakes, morphed images, chatbot personas and unauthorised merchandise. The order directs removal or disabling of identified material and provides for further similar content to be notified to the relevant defendants.
The underlying question. Whether Prabhu was entitled to interim protection against unauthorised uses of her identity, including AI-generated impersonation and commercial exploitation.
What this establishes. The relief is interim. The order preserves defendants’ ability to raise reasoned objections to removal requests and approach the court.
What happens next. The order lists the matter for 10 December 2026 and continues the ad-interim relief until further orders.
Source: Bombay High Court order, 1 October, paragraphs 19–25
Eighth Circuit grants interim relief in Minnesota synthetic image challenge
Jurisdiction: United States — federal / Minnesota
Court / authority: US Court of Appeals for the Eighth Circuit
Case / instrument: xAI challenge to Minnesota’s AI nudification prohibition
Area: Constitutional law · Synthetic intimate images · Interim relief
Stage and date: Appellate injunction reported · 2 October 2026
Previously in the report. Issue 007 covered xAI’s application for an injunction pending appeal. This week’s reported grant of interim relief advances that same matter. Read the earlier coverage.
What happened. Reuters, published by MPR News, reports that the Eighth Circuit granted xAI interim relief halting the law while its constitutional challenge continues. The district court had refused interim relief the previous month.
The underlying question. Whether xAI should receive interim protection from enforcement of Minnesota’s AI nudification law while its constitutional challenge proceeds.
What this establishes. The reported injunction pauses enforcement during the challenge; it is not a final determination that the law is unconstitutional.
What happens next. The constitutional challenge continues. Interim relief does not resolve the merits of xAI’s claim.
Source: Reuters via MPR News, 2 October
Google challenges EU access decisions and seeks interim relief
Jurisdiction: European Union
Court / authority: General Court of the European Union
Case / instrument: Google challenges to European Commission Digital Markets Act decisions
Area: Competition · Digital Markets Act · AI interoperability
Stage and date: Challenges and application for interim relief reported · 29–30 September 2026
What happened. Reuters reported Google’s challenges to two Commission decisions concerning rival access to search data and Android services available to Gemini. On 30 September, it reported that Google had also requested interim relief against the search-data order under Article 6(11) of the Digital Markets Act. Google cited privacy and security concerns; the Commission said its decisions contain safeguards.
The underlying question. Whether the Commission’s access requirements should be set aside and, separately, whether the search-data requirement should be suspended pending the court’s decision.
What this establishes. The interim application seeks suspension of the search-data requirement. The reported application is not a ruling granting suspension. Filing an action does not itself annul or suspend the Commission’s decisions.
Source: Reuters report, 29 September
Source: Reuters, 30 September: application for interim relief
Florida seeks temporary restrictions on OpenAI
Jurisdiction: United States — Florida
Court / authority: Circuit Court of the Tenth Judicial Circuit, Highlands County
Case / instrument: State of Florida proceedings against OpenAI and Sam Altman
Area: Consumer protection · Public nuisance · AI safety
Stage and date: Temporary-injunction motion reported filed · 28 September 2026
What happened. News4JAX and Law360 report that Florida’s attorney-general sought temporary restrictions on OpenAI’s model development without independent safety approval and on offering ChatGPT to minors in Florida. The motion also seeks limits on safety representations, human-like presentation and certain engagement practices in the existing lawsuit.
The underlying question. Whether Florida should obtain temporary restrictions on OpenAI’s activities and services while its consumer-protection and public-nuisance claims are litigated.
What this establishes. The requested restrictions are not a court order. Allegations of harmful or autonomous model behaviour remain allegations in this action.
Source: News4JAX report, 28 September
Source: Law360 report, 28 September
Regulators & Enforcement
Bank of England records increased frontier AI risks
Jurisdiction: United Kingdom
Court / authority: Bank of England Financial Policy Committee
Case / instrument: September Financial Policy Committee record
Area: Financial stability · Cyber risk · Operational resilience
Stage and date: Official risk assessment published · 30 September 2026 publication; meeting 25 September
What happened. The Committee said AI-related debt issuance had broadened capital-market exposure to AI developments. It also identified increased cyber and operational risks from frontier-model advances and recent test-environment incidents involving unexpected autonomous actions. It reiterated the need for firms to prepare for those risks and engage with relevant authorities and sector information-sharing work.
What this establishes. The record is not an enforcement decision against an AI provider or a finding that a particular firm breached a legal obligation.
Source: Financial Policy Committee record, September 2026
FTC confirms investigation into AI developers
Jurisdiction: United States
Court / authority: Federal Trade Commission
Case / instrument: Investigation concerning OpenAI, Anthropic and other AI companies
Area: Consumer protection · AI safety
Stage and date: Investigation publicly confirmed · 30 September 2026
What happened. The Associated Press reports that an FTC spokesperson confirmed an investigation into OpenAI, Anthropic and other AI companies over potential consumer risks. The spokesperson declined further comment.
What this establishes. Confirmation of an investigation is not an infringement finding. The reported announcement does not establish liability or impose a penalty.
Source: Associated Press report quoting FTC confirmation
California serves investigative subpoena on OpenAI
Jurisdiction: United States — California
Court / authority: California Department of Justice; Attorney General Rob Bonta
Case / instrument: Investigative subpoena in ongoing inquiry into OpenAI
Area: Enforcement · Cybersecurity · Frontier AI models
Stage and date: Subpoena served · 30 September 2026; announced 1 October
What happened. California’s attorney-general announced that an investigative subpoena had been served on OpenAI the previous day. The Department of Justice described it as part of a broader inquiry into cybersecurity incidents and risks involving the company and its models, following its previously announced investigation into the Hugging Face incident.
What this establishes. The subpoena is an investigative step. The announcement does not report a finding of liability, a penalty or a judgment against OpenAI.
Source: California Department of Justice announcement, 1 October
Information Commission succeeds the ICO
Jurisdiction: United Kingdom
Court / authority: Information Commission
Case / instrument: Governance transition under the Data (Use and Access) Act 2025
Area: Data protection · Regulatory governance
Stage and date: Functions transferred · 30 September 2026
What happened. The Information Commission took over the Information Commissioner’s functions. Statutory responsibility moved from a corporation sole to a body corporate with executive and non-executive members.
What this establishes. The government states that the transition does not change the regulator’s role, responsibilities or powers. It is an institutional change, rather than a new AI-specific compliance regime.
Source: Government announcement, 30 September
Legislation & Public Policy
California enacts rules for lawyers and arbitrators using generative AI
Jurisdiction: United States — California
Court / authority: Governor and California Legislature
Case / instrument: SB 574: attorneys, arbitrators, judicial officers and alternative resolution providers; 2026 Chapter 858; Business and Professions Code §6068.1; Code of Civil Procedure §§128.7 and 1282.1
Area: Professional responsibility · Court filings · Arbitration
Stage and date: Signed and chaptered · 30 September 2026
What happened. SB 574 bars delegating legal practice to generative AI. It requires reasonable verification and correction of outputs, disclosure of AI use in court-submitted documents, and personal verification of citations in covered court filings by a responsible submitting attorney, whether or not AI supplied them. It restricts entering confidential or nonpublic information into systems without appropriately restricted access. Arbitrators may not delegate decision-making to generative AI.
What this establishes. The law takes effect on 1 January 2027. It permits AI assistance while retaining professional responsibility. Its confidentiality restriction turns on access controls and obligations to protect the information.
Source: SB 574, chaptered text
Source: Morgan Lewis analysis, 2 October: commencement date
California signs broader AI legislation
Jurisdiction: United States — California
Court / authority: Governor and California Legislature
Case / instrument: SB 947, SB 951, SB 503, AB 1979, AB 2713, SB 1000 and SB 1111
Area: Employment · Healthcare · AI transparency · Digital replicas
Stage and date: Bills signed · 30 September 2026
What happened. The Governor signed measures addressing automated employment decisions and notice of technological displacement. The official announcement describes human review of disciplinary and termination decisions and disclosure of AI-related displacement. The package also includes healthcare, AI transparency and digital-replica legislation.
What this establishes. The announcement records enactment of separate measures. Their coverage and commencement depend on the individual statutes; the signing announcement does not make every provision immediately operative.
Source: Governor’s signing announcement and bill list, 30 September
Connecticut’s AI-related layoff disclosure takes effect
Jurisdiction: United States — Connecticut
Court / authority: Connecticut Legislature; Department of Labor
Case / instrument: Public Act 26-15, section 26
Area: Employment · AI disclosure · WARN notices
Stage and date: Disclosure requirement effective · 1 October 2026
What happened. Employers giving the Connecticut Department of Labor a notice required under the federal WARN Act must also disclose whether the layoffs relate to their use of AI or another technological change. The legislature’s research summary identifies 1 October as the effective date.
What this establishes. The provision adds information to an existing notice requirement. It does not create a universal notification requirement for every individual dismissal or prohibit AI-related layoffs.
Source: Connecticut Office of Legislative Research, 2026 Acts Affecting Business and Jobs
Source: Contemporaneous account of the provision taking effect, 1 October
White House accord sets out voluntary frontier model commitments
Jurisdiction: United States
Court / authority: President of the United States and participating technology companies
Case / instrument: White House Accord on Super Intelligence — Joint Commitment on Frontier Responsibilities
Area: Frontier models · Controls · Evaluation · Board oversight
Stage and date: Voluntary commitment published · 29 September 2026
What happened. The accord identifies four layers: internal controls during training and deployment; an internal team checking those controls; independent external audit or evaluation; and an independent board committee overseeing reports and remediation. Signatories represent Google, Anthropic, Meta, OpenAI, xAI and Nvidia. The text also commits participants to regular meetings on standards and practices.
What this establishes. The accord is not itself legislation or a regulatory enforcement decision. It contemplates possible future codification and does not establish that any signatory’s controls have passed independent assessment.
Source: White House accord, American Presidency Project
Executive order changes federal terminology for AI
Jurisdiction: United States
Court / authority: President of the United States
Case / instrument: Inaugurating the Era of Super Intelligence; Executive Order 14434
Area: Executive administration · AI terminology
Stage and date: Executive order issued · 29 September 2026
What happened. The order directs executive departments and agencies, to the maximum extent permitted by law, to use “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI” in specified official communications and non-statutory documents. For implementation, it adopts the existing statutory AI definition in 15 USC 9401(3). Within 60 days, the Assistant to the President for Science and Technology must submit proposed legislative language for a federal SI definition, including its relationship to the existing statutory AI definition.
What this establishes. The order does not require changes to previously issued regulations, presidential actions, contracts, grants or historical documents. Its terminology does not establish a new technical capability threshold.
Source: Executive Order 14434, White House
Brazil election restriction on new synthetic content begins
Jurisdiction: Brazil
Court / authority: Superior Electoral Court (TSE)
Case / instrument: 2026 electoral calendar and campaign-content rules; Resolution 23.760/2026; reference to Resolution 23.610/2019, Article 9-B
Area: Election law · Synthetic content · AI images and voices
Stage and date: Previously adopted restriction enters its first-round application period · 1 October 2026
What happened. The calendar prohibits publication, republication, including unpaid republication, and paid promotion of new AI-generated or equivalently altered synthetic content using the image, voice or expression of a candidate or public figure. The first-round period begins on 1 October and continues until 24 hours after the poll ends. The restriction applies even where the content is labelled and otherwise meets the referenced rules.
What this establishes. This is the application of an existing electoral rule, not a new judgment against a named platform or candidate, nor a prohibition on every use of AI. The second-round restriction begins on 22 October.
Source: TSE Resolution 23.760, consolidated electoral calendar
Government & Public Authorities
OpenAI and Anthropic decline attendance at scheduled Australian Senate hearing
Jurisdiction: Australia — federal
Court / authority: Australian Senate inquiry into AI and data centres
Case / instrument: Requests for company executives to attend
Area: Parliamentary scrutiny · AI safety
Stage and date: Attendance position reported · 28 September 2026
What happened. Reuters reported that OpenAI and Anthropic cited short notice for declining the hearing scheduled for 1 October. Written requests to their chief executives followed disclosures about an OpenAI agent’s unauthorised access to an Australian health-system database.
What this establishes. This was an advance attendance announcement, not a record of proceedings on 1 October.
Source: Reuters, 28 September: companies’ attendance positions
South Australia’s AI Royal Commission commences
Jurisdiction: Australia — South Australia
Court / authority: Royal Commission into Artificial Intelligence
Case / instrument: Royal Commission into Artificial Intelligence
Area: Public inquiry · AI governance · Workforce and public services
Stage and date: Inquiry commenced · 1 October 2026
What happened. The Royal Commission commenced with former Federal Court judge and economist Dr Iain Ross AO as chair, alongside Kate Pounder and Dr Christopher Manning. Its remit covers AI’s effects on the economy, industry, workforce and public services, and how South Australia can manage risks and build public trust.
What this establishes. The commission is an inquiry that will make recommendations. Its final report is due no later than 1 July 2027; its establishment does not itself impose new duties on AI developers.
Source: Royal Commission: commencement, commissioners and deadline
Source: Royal Commission: remit and engagement
Canada launches a National Council on Artificial Intelligence
Jurisdiction: Canada
Court / authority: Prime Minister of Canada
Case / instrument: Prime Minister’s National Council on Artificial Intelligence
Area: Public administration · AI policy · Safety and governance
Stage and date: Advisory council announced · 2 October 2026
What happened. Prime Minister Mark Carney announced a council bringing together public- and private-sector expertise to provide independent advice on Canada’s AI transformation. Its remit includes the evolution of the AI for All strategy, adoption, domestic companies, sovereign infrastructure, safety and democratic protection.
What this establishes. The announcement creates an advisory mechanism; it does not itself establish an AI enforcement regulator, impose penalties or enact new duties on private providers.
Source: Prime Minister’s announcement
Brazil outlines voice protection cooperation for the elections
Jurisdiction: Brazil
Court / authority: Superior Electoral Court (TSE)
Case / instrument: Cooperation with ElevenLabs
Area: Election information · Synthetic voice · Deepfakes
Stage and date: Official account published · 28 September 2026
What happened. The TSE described cooperation with ElevenLabs involving voice-based election information, a protected-voice database for designated public figures and identification of manipulated audio generated on the platform.
What this establishes. This is an account of technical cooperation. The TSE’s published account acknowledges that safeguards do not guarantee complete prevention of voice misuse.
Source: TSE announcement, 28 September
Next Dates
Brazil’s second-round synthetic-content restriction begins on 22 October. The Bombay High Court listed the Samantha Ruth Prabhu matter for 10 December, with interim relief continuing until further orders. California’s SB 574 takes effect on 1 January 2027. South Australia’s AI Royal Commission is due to report no later than 1 July 2027.
Sources and Scope
This edition covers developments dated 26 September–2 October 2026. Each entry links to its source and distinguishes court decisions and official instruments from authority announcements and attributed reporting.
The AI Law Report is an independent publication from AI Policies UK. It is provided for information and education and does not constitute legal advice or predict legal outcomes.
Next edition: Friday 9 October 2026.