Issue 7
Viewpoint
law · Hungary
Hungary adopts AI act and government decree, with fines of up to HUF 5.7 billion
Hungary has adopted its AI act and the accompanying government decree; fines of up to HUF 5.7 billion can be imposed in connection with high-risk AI systems. Meanwhile, Hungary's AI Act authority has appeared on the ministry's organisation chart.
Who is affected and how
- compliance specialist
- Assess which systems qualify as high-risk and adjust the compliance plan accordingly.
- corporate lawyer
- Legal risk arising from the fine levels must be managed in contracts and internal policies.
- lawyer
- Clients must be informed about the new Hungarian authority structure and the penalty regime.
per
British Columbia sues OpenAI and Sam Altman over Tumbler Ridge school shooting
British Columbia sued OpenAI and its chief executive, Sam Altman, in a US federal court on Monday over the school shooting in Tumbler Ridge that left eight people dead. The lawsuit alleges the…
lawyerLawyers should follow how the AI provider's duty to notify and warn develops in liability cases.corporate lawyerCorporate counsel should review liability insurance coverage and terms of use for AI providers and deployers.compliance specialistCompliance officers should build rules for flagging and escalating dangerous user content into internal procedures.regulation · European UnionDigital Omnibus: EU to delay application of AI Act high-risk rules
The Digital Omnibus on AI, handled in a separate fast-track procedure, would amend the AI Act's deadlines for high-risk systems, including the timing of the Annex III obligations. A main reason for…
compliance specialistCompliance roadmaps for high-risk systems must be re-planned around the amended deadlines, building in the uncertainty.corporate lawyerDelivery deadlines in AI procurement and supplier contracts should be aligned with the amended regulatory timetable.lawyerClients should be told that the delay is still part of a legislative process and is not final.ruling · United StatesAppeals court upholds Pentagon's designation of Anthropic as a supply chain risk
A federal appeals court in Washington ruled 2-1 that the government acted lawfully in designating Anthropic a supply chain risk. The ruling allows the Pentagon to continue removing Claude models from…
corporate lawyerThe risk of a supply chain risk designation should be reassessed in procurement and supplier contracts involving AI vendors.compliance specialistFor government clients, check which AI models could be blacklisted and align supplier compliance lists accordingly.lawyerLawyers handling government AI procurement should expect that the court has affirmed the government's designation powers.caseOpenAI agent accessed Australian government and health data without authorisation
Australia said an OpenAI agent accessed non-public government data, and the incident also affected a health database. Reports link the incident to a wider AI hacking campaign that may involve dozens…
compliance specialistAI agents' access rights and logging must be examined, because the incident involved agent access.corporate lawyerContracts involving AI agents need to clarify how liability for damage is shared.lawyerIn a data breach, the liability of both the AI agent's operator and the provider must be examined.Hungary
rulingHungarian court refers Gemini copyright case to EU Court of Justice (Like Company, C-250/25)lawyerIn Hungarian AI copyright cases, it is worth following how the preliminary ruling procedure develops.regulationHungarian universities' AI policies differ widely: cheating or help in higher educationprofessionNew term begins for award-winning AI course at Károli Reformed University's law schoolregulationHungarian ed-tech firms publish manifesto on AI use in schoolscitizenFor students and parents, the rules would set out what AI may be used for in schoolwork.
More key stories of the week
data protection · European Union
Digital Omnibus: EU member states split over AI and data transfer rules
According to MLex, EU governments are divided over the Digital Omnibus proposals on AI and data transfers. Privacy group noyb has published leaked Council documents on AI training and personal data…
compliance specialistGDPR compliance plans must account for possible relaxation of the rules on AI training and data transfers.lawyerInterpretations of the legal basis for processing personal data in AI training may change.citizenRelaxing the data rules affects the limits on using personal data in AI training.lawsuit · United StatesApple's $250 million Siri AI settlement opens for claims
The claims website for Apple's $250 million settlement over its Siri AI case has opened. People who bought an iPhone in late 2024 or early 2025 may be eligible for between $25 and $95; the case…
citizenAffected iPhone buyers can apply on the claims website to receive a share of the settlement.compliance specialistMarketing claims about AI capabilities must be documented and substantiated because of the risk of false advertising claims.corporate lawyerAdvertising AI features in products can lead to mass consumer lawsuits.regulation · United StatesGovernors, attorneys general, Bill Gates and House bills push for federal AI regulation
Governors from both parties, Minnesota's attorney general and Bill Gates are urging federal AI safeguards that go beyond industry self-regulation. Two bipartisan House bills would require…
corporate lawyerFederal AI regulation could reshape the compliance obligations of AI developers and deployers.compliance specialistA requirement for pre-deployment safety testing may call for a new internal review process.lawyerFollowing the legislative process helps in preparing clients.lawsuit · United States26 local publishers sue OpenAI and Microsoft over copyright infringement
Twenty-six local and trade news publishers filed a copyright lawsuit in federal court in New York on September 16 against the relevant Microsoft and OpenAI entities. The suit challenges the…
lawyerThe number of copyright lawsuits over AI keeps growing, so it is worth tracking the legal grounds of the claims.tool · globalOpenAI launches legal AI platform Astra for Law, with Sullivan & Cromwell and Cooley as early adopters
OpenAI unveiled a legal AI platform, Astra for Law, already used by firms including Sullivan & Cromwell and Cooley. Ropes & Gray built its own diligence tool together with OpenAI.
lawyerCompare the platform with existing legal AI tools, particularly for diligence and research work.corporate lawyerIn procurement, examine the data protection and confidentiality guarantees.ruling · United StatesFederal judge: Montana's deepfake election law likely unconstitutional
A federal judge has found that Montana's campaign advertising law on generative AI likely violates the First Amendment. The judge blocked the law but did not strike it down entirely.
lawyerFirst Amendment arguments are gaining strength in constitutional reviews of deepfake regulation.compliance specialistThe regulation of political AI advertising varies by state and remains uncertain.tool · globalLegora in talks at $8.5B valuation as Teachers' Venture Growth invests in Harvey
Legora is in talks to raise $300 million in fresh capital at an $8.5 billion valuation. Teachers' Venture Growth has invested in Harvey, the leading legal AI platform.
lawyerRapid fundraising by legal AI providers will speed up adoption of the tools in law firms.corporate lawyerWhen choosing a vendor, consider the provider's financial stability.lawsuit · United StatesUniversal and Warner ask court to dismiss musicians' AI licensing lawsuit
Universal and Warner have asked the court to dismiss the musicians' AI licensing lawsuit. The musicians are unhappy with the licensing agreements the labels have signed with AI platforms such as Suno…
lawyerThe interpretation of AI clauses in label–artist contracts may become the subject of legal disputes.regulation · United StatesNew York begins implementing RAISE Act: large AI developers must report safety incidents
New York State has begun implementing the RAISE Act: large AI developers must now register safety incidents with the state. Governor Hochul has signalled further regulatory steps.
compliance specialistLarge AI developers operating in New York must set up an incident registration process.corporate lawyerState-level AI obligations appear as a new item in compliance and contract frameworks.lawyerClients should be told that, in the absence of federal regulation, state rules become the benchmark.lawsuit · United StatesOpenAI, Anthropic, Google and SpaceXAI hit with antitrust class action over alleged AI slowdown pact
A class action has been filed in California against four major AI companies, alleging they unlawfully coordinated to slow AI development. The complaint says this could reduce the value of…
corporate lawyerThe competition-law risks of information sharing and cooperation between AI providers need to be assessed.compliance specialistSector-wide safety agreements require a competition-law review.lawyerIt offers a new class-action template for AI subscribers.profession · United StatesCourts punish hallucinated citations as US lawyers face patchwork of AI rules
US lawyers face differing court-by-court AI rules while courts sanction filings with fabricated citations. A Maryland lawyer was scolded again for fake citations after earlier AI-related problems.
lawyerCheck each court's own AI disclosure rules and verify every citation before filing.law · European UnionGermany and Italy implement the AI Act through national legislation
In Germany, the national KI-MIG statute implements the AI Act. In Italy, the topic is the reservation of human adjudication between Law No 132/2025 and the CEPEJ guidelines.
compliance specialistCompanies operating in several member states must check national implementation rules in each.lawyerDifferences in member state implementation matter for cross-border advice.In brief
professionLaw firms and clients clash over who reaps AI savings as in-house teams weigh freed capacitylawyerInstead of hourly billing, efficiency gains achieved with AI need to be handled transparently.corporate lawyerIt is worth fixing in advance, in engagement terms, how AI savings are shared.perUniversal and Sony file new copyright lawsuit against SunoprofessionFederal appeals court moves against AI-drafted filings after lawyers cite fake caseslawyerSources cited in any AI-drafted filing must be verified, because the court sanctions violations.regulationAnthropic and OpenAI ask Australia to ease its ban on AI training with local contentlawyerThe Australian direction of AI copyright regulation is worth following for international clients too.regulationRunaway AI agents spark debate over limits of European product liability lawcorporate lawyerWhen deploying AI agents, contractual liability and warranty terms need to be handled separately.lawyerThe applicability of the Product Liability Directive in cases of AI damage may generate legal disputes.compliance specialistRisk management must take into account liability arising from unexpected behaviour of agentic systems.lawStrict AI liability bills and expert calls to hold AI companies accountablecorporate lawyerGiven the strict-liability proposals, contractual exclusions between AI providers and deployers should be reviewed.lawyerNew grounds of liability may be invoked in AI-caused harm.data protectionGoogle fined €403 million over location data handlingcompliance specialistConsent and transparency practices for location data processing must be reviewed.corporate lawyerData protection fine risk should also be built into contractual and procurement risk analyses for large platforms.perFlorida State University shooting victim sues OpenAI, alleging ChatGPT helped plan the attacklawyerThe suit is another template for product liability and negligence claims against AI chatbots.corporate lawyerAI providers' risk assessments should treat claims arising from harmful use as a separate item.perUnsealed briefs: OpenAI and Microsoft executives knew book piracy was unlawful, authors saylawyerThe content of discovery between the parties may change the fair use and intent arguments.perUS Justice Department backs fair use for AI training in OpenAI copyright caselawyerIn fair use arguments, the government's filing may offer AI developers a new point of reference.caseHallucination-laced temporary restraining order raises alarm over courts' use of AIlawyerIf a court decision is flawed, options for legal remedies should be considered.data protectionEDPB issues new methodology for calculating GDPR finescompliance specialistFine-risk estimates must be reviewed under the new methodology.lawyerIn data protection authority proceedings, the fairness of a fine can be argued on the basis of the new methodology.data protectionUber's €825 million fine is a warning over automated dismissalscompliance specialistFor automated decisions affecting employees, human oversight and its documentation must be ensured.corporate lawyerHR system procurement and employment contracts should also set out the limits of automated decision-making.rulingNew Mexico jury finds Facebook executives misled users about their datalawyerIn consumer data protection cases, the misleading nature of executive statements can be proven.compliance specialistExecutives' statements on data handling must be verified and documented.regulationTaiwan's Judicial Yuan opens courts to Copilot AI usetoolLegal AI start-up Chamelio raises $26m Series A to replace legacy contract software with AI agentscorporate lawyerIn-house legal departments may assess AI-agent contract management systems as a replacement for traditional software.professionLegal AI company Harvey launches pro bono programme, with London leading the pushlawyerLawyers doing pro bono work can gain access to AI tools through the programme.professionCourts weigh AI use: African human rights court, Missouri and Wood CountyprofessionLaw firms and AI companies compete for legal technology talentlawyerThe legal engineer path offers a new professional alternative to the traditional legal career.caseTexas lieutenant governor candidate reports Dan Patrick's alleged AI-generated campaign material to policelawyerIn election deepfake cases, the relationship between the complaint and state rules needs to be examined.regulationSpain drafts AI rules as prime minister says self-regulation is not enoughcorporate lawyerSpain's regulatory direction could affect the compliance plans of companies across the EU.compliance specialistMember-state regulatory initiatives need to be monitored.professionReport outlines generative AI risks and safeguards for law officeslawyerOffice AI-use policies should be reviewed in light of the risks the report identifies.compliance specialistThe use of generative AI should be framed by internal rules and oversight.regulationOntario's new AI rules affect lawyers and expert witnesses as AI filings rise in Atlantic CanadalawyerOntario regulations must be understood before using AI.regulationNew York City and Oregon write their own frontier AI rules as Washington stallscompliance specialistAI developers operating in several states must expect differing local rules.corporate lawyerGiven the regulatory fragmentation, it is worth building a state-by-state risk map.complianceBanks warn AI shopping agents carry fraud and data privacy riskscompliance specialistFinancial institutions need separate fraud prevention and authorization controls for AI agent payments.corporate lawyerPayment contracts must clarify liability for transactions initiated by an agent.citizenPayment authorizations given to AI agents should be limited.professionUS law schools split between laptop bans and mandatory AI instructionregulationNvidia's Jensen Huang rejects antitrust and liability exemptions for AI labscorporate lawyerAI companies' requests for exemptions may influence the legislative debate on risk allocation.professionSouth Korean courts weigh tougher sanctions as AI hallucinations keep appearing in filingslawyerCheck sources in AI-assisted filings, as tougher sanctions may follow.professionAustralian court sanctions lawyer for improper use of generative AI in migration caselawyerMigration lawyers who use AI must check the accuracy of the material they file.toolCMS and Dentons sell stakes in legal AI company Noxtua to German legal publisherlawyerThe legal AI tools market is consolidating towards publishers, which affects the choice of tools.corporate lawyerThe impact of the change of provider on existing tool contracts needs to be examined.rulingDelhi High Court orders police protection for woman targeted with deepfake imagelawyerCourts can be asked for protective measures in deepfake cases.citizenVictims of deepfake images can seek court protection.perEmployers sued over AI monitoring and AI-recorded conversationscorporate lawyerA legal risk assessment is needed before workplace AI monitoring is introduced.compliance specialistAI recording without informing employees and obtaining their consent may be unlawful.lawyerA new type of employment and data protection litigation is emerging.data protectionIrish data protection authority publishes AI reportcompliance specialistThe report reflects the authority's data protection expectations on AI and can serve as a compliance baseline.rulingAyinde v Haringey: AI cannot hold a practising certificate or answer to a regulatorlawyerLawyers bear responsibility for their use of AI, because AI cannot be held professionally accountable.data protectionMeta fined for behavioral advertising in African jurisdictioncompliance specialistThe legal basis for personalized advertising must be assessed under local data protection rules.complianceAI compliance issues hit 40% of large companies in the past year, survey findscompliance specialistAudit trails must be built into AI processes, because traditional workflows do not cover AI.corporate lawyerCompliance gaps carry contractual and regulatory risk.regulationWorld's first binding international AI treaty stalls after signingcorporate lawyerThe convention's entry into force requires transposition by member states, which may affect compliance plans.caseHomeland security department turns to AI to help decide which public records are releasedcitizenUsing AI to assess freedom-of-information requests could reduce transparency.lawyerWhen appealing a refused records request, it is worth examining the use of AI as well.