Issue 8
Viewpoint
ruling · United States
Appeals court upholds Thomson Reuters win: AI training on its material is not fair use
On September 29 the US Court of Appeals for the Third Circuit upheld Thomson Reuters' (Westlaw) win against an AI competitor that used Westlaw headnotes for training. According to the sources, it is the first appellate ruling on AI training and fair use.
Who is affected and how
- lawyer
- In AI copyright cases, the ruling can serve as a starting point for arguments against fair use.
- corporate lawyer
- The contractual and data-source liability risks of AI developers and suppliers need to be reassessed.
- compliance specialist
- Documenting and verifying the provenance of training data has become more important.
regulation · European Union
EU AI Act after the omnibus: high-risk rules slip, transparency obligations do not
Analyses published in late September say the Digital Omnibus postpones the rules for high-risk systems to December 2027 and August 2028. Most transparency obligations remain, and according to the…
compliance specialistThe focus now should be on Article 50 labelling and GPAI obligations, not on the slipping high-risk deadlines.corporate lawyerAI labelling and transparency obligations should already be built into contracts and product descriptions.lawyerClients' AI Act classifications and the map of new deadlines should be updated.lawsuit · United StatesJustice Department backs OpenAI's fair-use defense in copyright case; news publishers object
On September 28 the US Department of Justice sided with OpenAI and Microsoft on the question of fair use in a copyright case. The news publishers in the case ask the court to give the government's…
lawyerFilings that rely on or attack a fair-use defense must take into account the differing positions of the Copyright Office and the Department of Justice.corporate lawyerGovernment support should be built into AI developers' legal risk assessments as a separate factor.law · United StatesCalifornia bans AI 'robo bosses': a human must decide on firings
Governor Gavin Newsom signed laws on September 30 restricting the use of AI in the workplace, after an earlier veto. According to reports, hiring and dismissal decisions must involve human…
corporate lawyerEmployers' AI tools must be reviewed to ensure that a human makes the dismissal decision.compliance specialistAI-based HR tools must be measured against the new requirements, and human review documented.lawyerFor clients who are California employers, compliance advice and new lines of defense in disputed dismissals open up.citizenCalifornia workers cannot be dismissed by an algorithm alone.lawsuit · United StatesFlorida attorney general seeks court order halting OpenAI's new models
On September 28, Florida Attorney General James Uthmeier asked the Highlands County 10th Judicial Circuit Court for a temporary injunction barring OpenAI and Sam Altman from developing new models…
corporate lawyerState-level restrictive measures could pose a risk to AI providers' operations.compliance specialistDocumenting and being able to prove independent safety testing matters if a state takes action.lawyerThe case tests how far states can go in restricting chatbots.Hungary
regulationHungary sets up Deputy State Secretariat for Artificial Intelligencecompliance specialistDomestic market surveillance coordination under the AI Act provides a new point of contact for compliance matters.corporate lawyerInformation on AI regulatory plans and legal harmonisation is best sought from the Deputy State Secretariat.lawyerIdentifying the competent body becomes easier when advising clients on the domestic application of the AI Act.data protectionHungarian data protection authority NAIH imposes record fine over AI analysis of customer service recordingscompliance specialistBefore AI analysis of voice recordings, the legal basis, information notices and data protection impact assessment should be reviewed.corporate lawyerWhen deploying AI tools in customer service, the NAIH decision must be built into the risk assessment.lawyerClients should be advised to undergo a data protection review for similar AI analytics projects.citizenAI analysis of customer service calls may be grounds for a data protection complaint.data protectionHungarian lawyer warns what data not to enter into AI chatbotslawyerClient data must not be entered into public AI chatbots, and the 72-hour deadline must be reckoned with in the event of an incident.compliance specialistThe chatbot-use policy and the incident-handling process must be designed around the 72-hour notification to NAIH.citizenIt is best not to enter personal and confidential data into AI chatbots.professionHungarian legal training on AI: business breakfast, MÜK lecture, 5 October workshoplawyerThe events cover practical questions of professional quality and AI labelling obligations, so they are worth attending.professionÓbuda University launches practical AI Act trainingcompliance specialistA domestic training opportunity opens up for the practical application of the AI Act.corporate lawyerA new opportunity is available for further training of a company's AI Act compliance team.toolOpten launches AI-based company information service, billed as Hungary's ChatGPT for company datalawyerA new AI tool is available for company information work, but its output must be checked against official data.corporate lawyerCompany checks can be done with faster data collection, provided accuracy is ensured.toolSlovak Hungarian-language statute database Törvénytár relaunches with AIlawyerAn AI-supported tool is available for searching Slovak legislation in Hungarian; its output must be checked against the official text.citizenThe Hungarian community in Slovakia gets easier access to legislation in force.
More key stories of the week
law · United States
California law regulates lawyers' and arbitrators' use of AI
Governor Gavin Newsom signed on September 30 the first California law, according to Bloomberg Law, to regulate the use of AI by lawyers and arbitrators. According to reports, responsibility for…
lawyerMaterials produced with generative AI must be checked paragraph by paragraph, and the check is best documented.corporate lawyerOutside counsel guidelines and engagement terms that touch on AI use should be reviewed.compliance specialistThe firm's internal AI policy and review process must be compared against the new requirements.lawsuit · United StatesAdvocacy group sues OpenAI over Hugging Face incident
An AI safety advocacy group filed a lawsuit against OpenAI on September 29 over the Hugging Face incident, arguing that the company is liable both as developer and as operator. According to a…
lawyerThe case is a test of developer and operator liability when AI agents cause harm.corporate lawyerWhen deploying agentic AI, the supplier contract must clarify the allocation of liability and incident handling.compliance specialistAccess rights and logging for AI agents should be re-examined in light of the leak risk.citizenIf user data and images are leaked, claims for damages and data protection remedies may arise.profession · United StatesLawdistrict analysis finds 574 AI-hallucinated court cases; Florida warns of tougher sanctions
A Lawdistrict analysis found 574 court cases invented by AI. A Florida appellate court warned that lawyers who rely on AI recklessly face increasingly severe sanctions.
lawyerEvery AI-generated citation must be checked before a filing is submitted, because sanctions are getting heavier.case · United StatesAnthropic says rogue AI agents pose uncertain legal risk
Reuters reported on September 29 that Anthropic says customers and users could sue the company over the actions of AI agents, while the legal framework remains uncertain. According to the Hungarian…
corporate lawyerContracts for agent-based products should include limitations of liability and user oversight obligations.compliance specialistThe risk register and risk disclosures covering agents' actions are worth expanding.lawyerFor harm caused by an AI agent, one has to choose between liability theories, as the legal framework has not settled.tool · globalClio acquires Learned Hand, an AI startup built for judges
Legal software company Clio acquired Learned Hand on September 30; according to reports, the startup builds an AI workspace for judges and law clerks that connects to case management systems. The…
lawyerExpect AI support to spread to the bench as well, which may also affect how arguments and filings are shaped.ruling · United StatesArizona appeals court: judge wrongly allowed AI-generated victim video
According to September 30 reports, an Arizona court found that the judge wrongly allowed an AI-generated victim video to be shown. The decision bears on the limits of using AI content in criminal…
lawyerCiting the decision, counsel can object to such videos being shown, for the defense and the prosecution alike.regulation · United StatesPolitical deepfakes: lawmakers fall behind as campaigns turn to legal action
As AI-generated political deepfakes spread, lawmakers are struggling to keep pace, with 31 states regulating election deepfakes ahead of the midterm elections. The Cooke campaign sent a…
lawyerIn election deepfake cases, counsel should build on state-law grounds and the bad-faith standard.compliance specialistPolitical campaigns' AI content and distribution controls must be managed under changing state rules.citizenSteps can be taken over fake videos on the basis of defamation.law · United StatesConnecticut AI and data privacy laws take effect October 1
New AI and data privacy laws take effect in Connecticut on October 1, 2026. The debate among experts is whether the state is writing the regulation that large companies want.
compliance specialistCompanies working with Connecticut customers must update their data handling and AI processes by the effective date.corporate lawyerContractual liability and data handling clauses need to be compared with the new state rules.law · United StatesSenators Hawley and Murphy announce bipartisan AI Agent Accountability Act
Senators Josh Hawley and Chris Murphy announced the bipartisan AI Agent Accountability Act on October 1. The Senate is debating liability for harm caused by rogue AI agents.
corporate lawyerThe bill could strengthen the direct liability of companies that operate AI agents, so it is worth assessing contractual risk now.lawyerResponsibility for agents' activities may emerge as a new legal basis in liability cases.compliance specialistThe bill should be tracked for its implications for governance frameworks for agent deployments.regulation · United StatesTrump calls AI self-regulation important; no consensus reached at White House
After a meeting with technology leaders on September 29, Donald Trump called AI self-regulation "very important". He said no agreement emerged on guardrails.
compliance specialistMandatory federal regulation is unlikely to arrive soon, so state rules and voluntary frameworks need to be tracked.corporate lawyerState and federal regulatory uncertainty should be factored into contractual risk management.ruling · United StatesJudge dismisses Chegg and Penske antitrust lawsuits over Google's AI search
According to Ars Technica, a court on 1 October dismissed the antitrust complaints filed by Chegg and Penske against Google's AI-based search. The court acknowledged the consequences of AI search but…
lawyerContent publishers' competition-law arguments in AI search cases have hit a serious obstacle, so copyright or other legal grounds may be more effective.corporate lawyerUnder the ruling, the antitrust risk for platforms over AI search is limited.lawsuit · United StatesSuno fails to escape music copyright suit as labels file new complaint
A US court has held that independent artists need not identify specific infringing outputs for their copyright claims against Suno to proceed. Universal and Sony sued Suno again in September over its…
lawyerThe pleading threshold is more favourable in suits against AI music generators.In brief
professionAI-native law firms launch as Arceus Legal raises $17MlawyerFixed-fee offers from AI-native firms may put price pressure on traditional hourly-billed services.corporate lawyerAI-based firms may also become competitors when choosing external legal service providers.caseAuthors' groups urge publishers to give up Anthropic settlement claims on out-of-print bookslawyerSeparate contractual arguments should be prepared for distribution disputes between authors and publishers.lawSenate AI safety proposals: transparency bill, ban on recursive AI, blocked fast-track votecompliance specialistGiven the federal bills, internal AI safety and transparency controls should be assessed in good time.corporate lawyerBecause federal regulation is uncertain, product development contracts need a reserve for legislative changes.lawNewsom signs 11 more AI safety laws as his term endscompliance specialistAI companies operating in the California market need to review the obligations that are newly taking effect.corporate lawyerThe compliance map needs to be updated in line with the new California laws.regulation26 state attorneys general urge Congress to regulate frontier AI while preserving state authoritycorporate lawyerIf state AI laws are preserved, companies will need to prepare for compliance under several legal regimes.compliance specialistFederal preemptive legislation could decide whether state requirements remain in force.regulationEU states urged to go beyond the AI Act to protect people from abusive surveillancecitizenStronger protection at member-state level could improve people's ability to enforce their rights against surveillance.compliance specialistFor AI systems used for surveillance, additional national rules must also be monitored.rulingAppeals court temporarily blocks Minnesota's AI 'nudification' law in xAI suitlawyerConstitutional challenges to state deepfake laws can succeed, which matters for legal arguments.compliance specialistThe validity and suspension of state requirements must be tracked.citizenLegal protection against deepfakes at state level may weaken temporarily.caseHallucination-laced court order raises concerns in MinnesotalawyerSpotting such errors can also be useful when reviewing the opposing party's reasoning.data protectionIreland's DPC reports significant rise in AI-related engagementscompliance specialistData protection impact assessments for AI deployments should be done in advance, because agentic cases are set to increase.corporate lawyerFor AI developments on Ireland-based platforms, prior consultation with the DPC may often be expected.perElon Musk summoned over Grok deepfake lawsuitslawyerIn deepfake victims' lawsuits, both the platform and the model developer can be held liable.compliance specialistContent-filtering obligations for generative image services are becoming stricter.citizenIn the case of a fake sexual deepfake, legal action can be taken against the service provider.rulingWuhan court issues final ruling on unauthorized copying of AI-generated mini-serieslawyerThe Chinese ruling can serve as a reference in comparative studies of the protection of AI works.rulingTokyo court extends legal protection to voice against AI cloneslawyerThe Japanese court's direction can be cited in voice-cloning cases.rulingMunich court extends AI copyright protection in GermanylawyerThe Munich ruling should be monitored in cross-border copyright cases.regulationEU reopens the copyright file in the age of AIperOpenAI cites free speech in seeking dismissal of lawsuit linked to FSU shootinglawyerFree-speech protection for chatbot output may become a key issue in liability lawsuits.corporate lawyerBoth product liability and free-speech defenses matter for the risk faced by AI providers.rulingDC Circuit upholds Defense Department's exclusion of Anthropic's AI from supply chaincorporate lawyerJudicial review of supply-chain risk exclusions in government contracts may be narrow.lawyerClient risk from government exclusion of AI vendors should be assessed in light of the decision.perCourt filings show OpenAI knew training AI on pirated books was illegallawyerIn copyright cases, disclosure of internal communications can be key evidence.regulationCSBS announces AI supervisory framework for state-chartered banks and nonbank financial institutionscompliance specialistUnder the supervisory framework, financial institutions' use of AI must be documented and controlled.corporate lawyerAI vendor contracts of financial service providers must be aligned with supervisory expectations.regulationNew Mexico prepares one of the strictest AI regulatory pushes for 2027compliance specialistAI developers that also operate in New Mexico need to assess the strict state expectations in advance.corporate lawyerThe proposed state rules could also affect developer liability and product-safety commitments.regulationAI training and copyright debate reaches a threshold moment for India's media industrylawPennsylvania House passes bill regulating AI chatbotscompliance specialistCompanies operating chatbots should follow the bill's further progress.corporate lawyerUser interfaces and liability disclaimers may have to change because of the state requirements.regulationSingapore launches public consultation on AI and intellectual property lawlawyerClients in the region working on AI and IP should be informed about the rules now being drafted.professionIsrael Bar Association issues expanded ethics opinion on lawyers' use of AIlawyerWhere there is an Israeli connection, AI use must be documented and checked in line with the ethics rules.compliance specialistA law firm's AI policy should be compared with the Bar Association's position.toolIvo open-sources Sage, a legal AI model for long-horizon contract workcorporate lawyerFor contract review, a free model that can be run in-house may be considered, subject to a data protection check.lawyerThe model's output must be checked, and data security conditions must be examined separately.complianceAI deepfake fraud raises liability stakes for banks and businessescompliance specialistVerification controls for voice- and video-based identification should be strengthened because of deepfake risk.corporate lawyerThe contractual and liability rules on who bears fraud losses should be reviewed.rulingDelhi High Court orders takedown of deepfakes targeting Drishti IAS founderlawyerAn immediate takedown request based on personality rights can succeed against AI deepfakes too.citizenCourts can be asked to order the quick removal of deepfakes.caseOne of autumn's biggest French book hits dropped from Goncourt list over suspected AI usecomplianceNew York attorney general urges AI company employees to report wrongdoingcompliance specialistAI companies must provide an internal whistleblowing channel and protection against retaliation.corporate lawyerThe risk of state enforcement rises if whistleblower protections are violated.caseTwo-lawyer Australian firm wins tax case with argument devised by AIlawyerSmaller firms can also use AI to develop arguments, provided they verify the result themselves.professionTouro Law introduces mandatory AI training for incoming studentstoolArbitrus unveils AI 'judge' designed to resolve legal disputeslawyerBefore accepting AI-based dispute resolution, the parties' consent and review of the output must be clarified.data protectionReport: Copilot privacy practices allow humans to view uploaded photoscompliance specialistBefore Copilot is used at work, the provider's terms on human review should be checked.citizenThe terms of service should be checked for whether uploaded photos may be reviewed by humans.