Issue 4
Viewpoint
regulation · European Union
EU AI Act: Digital Omnibus delays high-risk deadline, but obligations remain in force
The deadline for high-risk systems has been postponed through the Digital Omnibus amendment of the AI Act, and the change is now law, not a proposal; after the provisional agreement of 7 May, however, many obligations still apply. Analysts say the delay is not an amnesty, the EU has begun enforcement, and big tech companies are already adapting.
Who is affected and how
- compliance specialist
- Despite the delay, obligations already in force (such as transparency) must be met on time, and preparation for high-risk systems must continue.
- corporate lawyer
- The roles under the AI Act (provider, deployer) and the Annex III classification need to be reviewed, including for contracts.
- lawyer
- Clients need to be told what has been postponed and what remains in force.
data protection · European Union
Dutch regulator fines Uber €824.99m over automated driver suspensions
The Dutch Data Protection Authority (AP) fined Uber €824.99m for automatically suspending drivers without human review between 2018 and 2022. The company has appealed; the penalty is the…
compliance specialistCompanies must assess where they make decisions with legal effect without human review and build human intervention in at those points.corporate lawyerThe GDPR risk of algorithmic decisions about workers and partners, and the likelihood of regulatory action, should be reassessed.lawyerThe decision gives clients who challenge automated account suspension or dismissal a new basis to cite.lawsuit · United StatesAnthropic's $1.5 billion settlement: pirated books, not training, triggered the payment
Anthropic's $1.5 billion copyright settlement is the first major AI settlement to receive court approval, and the first from which authors can claim money, at about $3,000 per book. Commentators say…
corporate lawyerFor clients developing models, the provenance of training data should be audited, because infringing acquisition is an independent basis for liability.lawyerClients should be advised on the distinction between fair use and pirated sources.ruling · United StatesJudge rules Pentagon and administration measures against Anthropic were unlawful
A federal judge ruled late Thursday that the Trump administration unlawfully penalized Anthropic, and permanently barred the application of rules aimed at excluding the company from the federal…
corporate lawyerLegal risk assessments for AI companies with government supplier relationships must take into account that a court can overturn bans.compliance specialistIn reviewing federal procurement and supplier compliance risks, the ruling shows the limits of blacklisting measures.lawyerThe case can serve as an example of legal protection against government exclusions.ruling · European UnionMunich court: creating music with generative AI may be copyright infringement (GEMA v Suno)
The Munich Regional Court ruled in GEMA's lawsuit against Suno that creating music with generative AI may constitute copyright infringement. DLA Piper has published an analysis of the decision.
corporate lawyerMusic AI providers operating in Europe need to reassess their output and training infringement risk.Hungary
perHungarian local media's lawsuit against Google: EU Court of Justice to rule on AI training for the first timecorporate lawyerFor clients developing or using AI in the EU, the training-data risk assessment should be aligned with the judgment.lawyerIn Hungarian copyright practice, the EU ruling may provide guidance for similar claims.complianceHungarian AI Act guides: KPMG, SME checklist, lawyer interview, Blikk seriescompliance specialistSMEs should start AI Act compliance by assessing their AI tools, roles and risks.corporate lawyerFor Hungarian clients, classifying roles and identifying missing controls is the practical first step.lawyerColleagues with Hungarian AI law specialist training are increasingly sought after in AI Act advice.data protectionPoland investigates ChatGPT over alleged data privacy breachcompliance specialistCompanies that use ChatGPT should re-examine their data controller responsibilities under the GDPR.citizenAffected individuals can file a data protection complaint if their data was involved in the incident.caseProArt: AI puts nearly a fifth of Hungarian music royalty revenue at risklawSerbia's draft data protection law: controllers and processors have until 10 September to preparecompliance specialistControllers operating on the Serbian market must prepare for the new rules by 10 September.corporate lawyerThe overlap between the EU AI Act and Serbian rules must be addressed in the compliance programme.toolWolters Kluwer builds structured legal knowledge base to develop Libra AIlawyerIn Hungarian legal research, Libra AI's answers should be used with source verification.
More key stories of the week
tool · global
Cleary Gottlieb partners with Google Cloud to deploy Gemini Enterprise for Legal
Cleary Gottlieb has partnered with Google Cloud to deploy Gemini Enterprise for Legal. According to Golem.de, Google is aiming the offering at law firms.
lawyerAnother major vendor has joined Harvey and Thomson Reuters in firms' tool selection.corporate lawyerWhen procuring cloud-based legal AI, data handling and confidentiality terms should be requested.tool · globalThomson Reuters launches proprietary LLM trained on Westlaw and Practical Law content
Thomson Reuters has launched its own language model, "Thomson", trained on Westlaw and Practical Law content, challenging Harvey and Legora. According to Startup Fortune it is built on Alibaba's Qwen…
lawyerLegal research tools should be compared on benchmarks and on the firm's own workflows, because the models lead in different areas.corporate lawyerWhen switching vendors, the model's origin (a Chinese base model) should also be examined.lawsuit · United StatesSony and Warner music publishers sue Anthropic over pirated music works
Sony and Warner music publishers have sued Anthropic, alleging "blatant theft" and seeking damages of up to hundreds of thousands of dollars per work, claiming the company pirated musical works…
corporate lawyerAI providers must document their training data sources, because the risk of per-work damages is high.regulation · United StatesArizona Supreme Court declines to ban judges' AI use; New Mexico drafts statewide court guidelines
The Arizona Supreme Court rejected a petition to ban judges from using any AI and will continue its controlled testing of the technology in the court system. New Mexico judges are drafting statewide…
lawyerRules on AI use in courts affect how filings and evidence are handled.regulation · United StatesUS AI regulation: 60+ state rules, California amendment, dispute over federal preemption
According to a Taft compilation, more than 60 state-level AI rules apply to commercial uses, from health-care notices to chatbot disclosures. The California Assembly passed amendments to the AI…
compliance specialistA matrix of state-level rules must be maintained, because requirements differ for health-care notices and chatbot disclosures.corporate lawyerFor clients operating in several states, the California amendments and the federal preemption debate need to be followed.lawsuit · United StatesWikiHow sues OpenAI for copyright infringement over AI training
WikiHow has sued OpenAI for copyright infringement over AI training. The news appeared earlier this week.
corporate lawyerFor clients building AI, the legal risk of data collected from public websites should be reassessed.caseBrazil court sanctions lawyers for hidden AI prompt in pleading
Brazilian courts have sanctioned or investigated lawyers who placed hidden AI prompts in their filings. According to an analysis by a Paris lawyer, French lawyers also need to check their filings…
lawyerFilings should be checked for hidden text before submission, as sanctions are possible.profession · globalYoung lawyers and AI: automation threatens the entry level
At the SIAC symposium, panellists discussed the impact of AI on junior lawyers in arbitration; Bill Gates said white-collar jobs, including paralegal work, are affected. Professional articles debate…
lawyerFirms must rethink their training model for juniors.professionAI hasn't triggered flood of legal malpractice claims, but insurers are watching
The number of legal malpractice claims has not risen because of AI, but insurers are monitoring the area. Insurers have given law firms five recommendations for reducing AI risk.
lawyerFirm-wide AI use policies and verification processes should be aligned with insurers' expectations.corporate lawyerProfessional liability policy terms should be reviewed for exclusions related to AI use.lawsuit · United StatesUS judge skeptical of Meta workers' AI layoff claims, points to arbitration
A US judge expressed skepticism about Meta workers' AI-related layoff claims and pointed to arbitration proceedings. MLex reported the case.
corporate lawyerArbitration clauses in employment contracts can offer protection against AI layoff claims.lawyerOn the employee side, a strategy is needed to get around arbitration clauses.ruling · United StatesJudge allows UMG and Sony to pursue Suno over YouTube stream-ripping and output claims
A US judge has allowed UMG and Sony to accuse Suno of pirating from YouTube through "stream ripping" to train its AI. The judge also dismissed an independent artist's class action, so claims…
lawyerIn discovery, documents on Suno's training data sources can be requested.compliance · European UnionAI Act Article 50: chatbots must be labelled, but most Dutch companies do not do it
Article 50 of the AI Act has applied since 2 August: when people talk to AI, this must be disclosed at the point of interaction, in their language. A Dutch survey found that only 14 of 100 companies…
compliance specialistFor customer service chatbots, the placement and language of the AI label must be checked, because practice falls short of the requirement.corporate lawyerContracts with chatbot providers should build in fulfilment of the transparency obligation.In brief
caseMore sanctions for fake AI-generated citations in US courtslawyerEvery AI-assisted citation must be checked before filing, as sanctions can fall on both the party and the lawyer.caseWho is liable when an AI agent goes rogue? Survey of counsel points to rising dispute riskcorporate lawyerBefore deploying AI agents, the liability chain (developer, supplier, deployer) should be set out in contract.compliance specialistThe permissions and oversight of autonomous agents must be documented.data protectionAlabama opens probe into OpenAI after Hugging Face breachcompliance specialistCompanies that use model-hosting and data-sharing platforms must review third-party data security risks.corporate lawyerCompanies should also begin preparing for state attorney general investigations.perSurvivors of sexual abuse sue xAI over Grok trainingcorporate lawyerAI providers should treat the duty to filter training data and generated content as a liability risk.citizenVictims can pursue claims through a class action.caseUS appeals court weighs reassigning case from judge over AI uselawyerAI authorship of a judicial decision may become grounds for a procedural objection.perAI hiring litigation: lessons from Mobley v. Workday for employerscorporate lawyerWhen using AI-based hiring tools, discrimination risk must be addressed in the vendor contract and internal controls.compliance specialistRecruiting algorithms should be audited regularly.caseSecrets confided to ChatGPT have surfaced in court casescitizenPersonal information typed into AI chats can be used in litigation.lawyerClients should be told that their chatbot conversations can become evidence and are not protected by confidentiality.rulingIrish High Court clarifies DSAR exemptions in TikTok appeal against data protection regulatorcompliance specialistIn DSAR handling procedures, exemption grounds must be applied as set out in the ruling.lawyerThe Irish ruling can serve as a reference point in disputes over data subject access.rulingHamburg court restricts xAI from spreading false information through AIcorporate lawyerNational courts can also restrict the content liability of European AI providers.lawyerPersonality-rights claims over incorrect AI output can be brought.perCourt lets privacy claims against Otter.ai proceedcompliance specialistWhen AI note-takers are used, participants' consent must be obtained.lawyerAn AI note-taker used in client meetings can breach confidentiality.professionJudges worldwide: AI may decide minor cases, but judicial conscience cannot be replacedcitizenLitigants without a lawyer must check filings they prepare with AI.perArgentina's Perfil sues OpenAI and Microsoft over use of journalistic contentlawyerFor publisher clients, a claim on multiple legal grounds can be considered.caseColorado investigates possible deepfake-law violation in election videocompliance specialistCampaign materials must be checked for AI-use labelling under state disclosure rules.citizenUnder the disclosure requirement for the AI origin of election videos, a complaint can be filed.professionLegal AI startups raise more funding: Newcode $13.5m, Discernis $2.5mcorporate lawyerVendor risk and data protection due diligence is essential for new startups.toolNew agentic features for law firms: CoCounsel, Aderant, EverlawlawyerResponsibility for oversight must be clarified before agents are used in litigation and investigation workflows.perTwitch streamer sues Amazon over use of streamers' content to train AIcorporate lawyerPlatforms should review whether the training permissions in their user terms are clear.perCarl Sagan's estate sues Luma over use of his voice in an AI adtoolLexisNexis unveils agentic Protégé and plans its own AI modellawyerResults from agentic research features must still be verified when trying them out.toolCasetext cofounder Pablo Arredondo joins Clio to lead its judiciary expansionregulationUK Jurisdiction Taskforce consults on liability for AI harmscorporate lawyerLiability allocation in contracts with AI providers should be aligned with developments in English law.lawyerProfessional responsibility remains with the professional when AI is used.regulationKorean music copyright group withdraws AI-assisted song registration ruletoolJackson Lewis and Nelson Mullins announce firmwide Harvey rolloutslawyerA firmwide rollout requires training and internal usage rules.caseConnecticut man hid AI prompt in court filinglawCalifornia SB 574: will AI's home state kill off AI for law?lawyerCalifornia lawyers should watch the bill, as it may affect the legal framework for AI use.rulingFact check: did a judge rule some AI-generated abuse material is protected by the First Amendment?lawyerIn criminal defence, the treatment of AI-generated material must be assessed against the exact text of the ruling.regulationSingapore opens public consultation on copyright and AI trainingregulationNew state laws protect users of AI chatbots used as therapists and companionscompliance specialistFor mental health chatbots, state requirements need to be matched with compliance practice.citizenUsers have legal protection when they use AI chatbots for therapy.caseDutch lawyer fined €2,300 for AI-hallucinated citationlawyerA cited ruling must always be checked against the source, as an error can bring a fine.complianceDutch regulator outlines process for AI Act fundamental rights impact assessment (FRIA)compliance specialistDeployers of high-risk systems must align their impact assessment with the regulator's process.rulingBrazil's electoral court to rule next week on AI Bolsonaro avatarcompliance specialistRules on the use of AI avatars in election campaigns may offer guidance.lawNew Louisiana laws address school threats and misuse of AI-generated imagescitizenUnder the new laws, misuse of AI-generated images can be penalised.perAmazon–Perplexity lawsuit could shape the law on AI shopping agentscorporate lawyerContract terms on AI agents' access to platforms need to be reviewed.complianceAz SEC megegyezett egy miami AI-startuppalcompliance specialistThe accuracy of investor and marketing claims about AI capabilities must be ensured.regulationAI floods public bodies with complaints and information requestscompliance specialistPublic bodies need to set up a procedure for handling mass requests generated with AI.citizenRequests prepared with AI must also meet the procedural rules.lawGermany's AI migration act raises concerns over safety, fairness and legal standardslawyerChallenging AI-assisted decisions in migration cases raises new legal questions.citizenPeople affected by AI-assisted migration decisions need access to a legal remedy.