Burhan Doğuş Ayparlar

Case database

AI decisions

Court and regulator decisions on artificial intelligence from Türkiye, Korea and around the world, each summarised as facts, question, decision and significance, in three languages.

46 decisions shown

  1. South Korea

    Seoul Administrative Court · Kakao Pay v. Personal Information Protection Commission

    Court upholds penalty on Kakao Pay for sending users' data to Alipay for risk scoring

    Seoul Administrative Court treated Kakao Pay's transfer of some 40 million users' data to Alipay for NSF scoring as an unconsented third-party provision.

    Under appealData protectionAutomated decisions
  2. South Korea

    Patent Court of Korea · Company A (anonymised) invalidation action

    Patent Court records in its judgment that counsel filed non-existent, AI-generated precedents

    In an invalidation case the Patent Court stated that many precedents the plaintiff relied on did not exist and that filing them unverified was inappropriate.

    FinalFabricated citations
  3. Türkiye

    Ankara Regional Court of Appeal, 20th Civil Chamber · Yapay zekâ ile ses klonlama (görev uyuşmazlığı)

    AI voice-cloning claim belongs in the general civil court rather than the IP court

    The chamber held that a claim over unauthorised AI use of a person's voice is a personality-rights and data-protection dispute, not a copyright one.

    FinalDeepfakesPersonality rightsData protection
  4. Türkiye

    Turkish Competition Board · Meta Platforms, Inc., Meta Platforms Ireland Limited, WhatsApp LLC ve Meta Platforms İstanbul Bilişim Hizmetleri Ltd. Şti.

    Meta ordered, as an interim measure, to open WhatsApp to rival AI chatbots

    The Board opened an investigation into Meta over Meta AI in WhatsApp and ordered it to let rival generative AI assistants operate there.

    Interim rulingCompetition
  5. Türkiye

    Istanbul 14th Commercial Court of First Instance · Tanıma ve tenfiz (Amsterdam Mahkemesi kararı)

    Commercial court discloses in its judgment that it used AI to verify and translate foreign case law

    In an enforcement case the court recorded that it used AI only as a research and translation aid, under the Public Officials Ethics Board principles.

    Under appealTransparencyPublic sector
  6. Türkiye

    Personal Data Protection Board (KVKK) · Mesai Takibi Amacıyla Biyometrik Veri İşlenmesi Hakkında İlke Kararı

    Fingerprint and facial recognition attendance systems held unlawful even with employee consent

    By a principle decision the Board found biometric attendance tracking disproportionate and required cards, PINs or signatures instead.

    FinalData protectionBiometrics
  7. Italy

    Tribunale di Roma (Rome Civil Court) · OpenAI v Garante per la protezione dei dati personali (ChatGPT)

    Italian DPA's EUR 15 million ChatGPT fine against OpenAI annulled for lack of competence

    The Rome court annulled the Garante's fine, holding that competence passed to the Irish authority once OpenAI had a main establishment in Ireland.

    Proceedings ongoingData protectionChatbot liability
  8. South Korea

    Constitutional Court of Korea · Transfer of immigration face data for AI training (2021Hun-Ma1575 et al.)

    Constitutional Court dismisses complaint over use of immigration face data to train AI

    The Court unanimously dismissed as inadmissible a complaint over some 170 million immigration face images made available to firms to train AI.

    FinalBiometricsPublic sectorData protection
  9. United Kingdom

    UK Supreme Court · Emotional Perception AI Ltd v Comptroller-General of Patents, Designs and Trade Marks

    Supreme Court abandons the Aerotel test and holds an ANN-based invention is not excluded as a program 'as such'

    Adopting the EPO's 'any hardware' approach, the Supreme Court overturned the refusal of an ANN-based application and sent it back for examination.

    FinalPatents
  10. Australia

    Administrative Review Tribunal (Guidance and Appeals Panel) · Bunnings Group Limited and Privacy Commissioner

    Retail facial recognition fell within a “permitted general situation”, but notice breaches were upheld

    The tribunal held that Bunnings could rely on a consent exception for its facial recognition system but upheld findings on notice and governance.

    FinalBiometricsData protection
  11. Germany

    Hanseatic Higher Regional Court of Hamburg (OLG Hamburg) · Kneschke v LAION e.V.

    Downloading photos to build an AI training dataset held covered by the text and data mining exceptions

    OLG Hamburg held that LAION's copying of a photo to build a dataset was covered by §§ 44b and 60d UrhG; the case is now before the Federal Court of Justice.

    Under appealCopyright and training
  12. Germany

    Munich Regional Court I (Landgericht München I), 42nd Civil Chamber · GEMA v OpenAI

    Memorisation of song lyrics in a language model held to be reproduction; OpenAI found liable

    Munich Regional Court I held that memorising lyrics in ChatGPT's models and reproducing them in outputs infringed copyright; the TDM exception did not apply.

    Under appealCopyright and training
  13. United Kingdom

    High Court of England and Wales, Business and Property Courts (Intellectual Property List) · Getty Images (US) Inc & Ors v Stability AI Ltd

    Stable Diffusion held not to be an 'infringing copy' because it does not contain the training works

    The High Court rejected Getty's secondary copyright claim over Stable Diffusion and found only limited, historic trade mark infringement from watermarks.

    Under appealCopyright and training
  14. South Korea

    Supreme Court of Korea · Naver v. Korea Fair Trade Commission

    Supreme Court sets aside ruling against Naver over search-algorithm self-preferencing

    The Supreme Court set aside a judgment upholding a KRW 26.6 billion fine on Naver and required concrete proof of anticompetitive effect and intent.

    Proceedings ongoingCompetitionConsumer protection
  15. South Korea

    Supreme Court of Korea · Supreme Court 2024Do17801 (anonymised)

    Supreme Court sets the test for when deepfakes of a real minor count as child sexual exploitation material

    The Supreme Court held that deepfakes using a real minor's face can fall under the “clearly recognisable representation” limb when they depict sexual acts.

    FinalDeepfakesCriminal law
  16. Türkiye

    Ankara 7th Criminal Judgeship of Peace · Grok (X / xAI) paylaşımları

    Access blocked in Türkiye to 50 posts generated by the Grok chatbot

    After prosecutors opened a probe into Grok's insulting replies, a criminal judgeship of peace blocked access to 50 of its posts.

    Interim rulingChatbot liabilityCriminal law
  17. United States

    U.S. District Court for the Northern District of California · Bartz v. Anthropic PBC

    Training on books held fair use, but building a library from pirated copies was not

    The court held that training Claude on books was fair use but downloading pirated copies for a central library was not; the case then settled.

    SettledCopyright and training
  18. United Kingdom

    High Court of England and Wales, King's Bench Division (Divisional Court) · R (Ayinde) v London Borough of Haringey; Al-Haroun v Qatar National Bank QPSC and QNB Capital LLC

    Lawyers who put fictitious, AI-generated authorities before the court referred to their regulators

    The Divisional Court held that lawyers must verify AI-assisted research against authoritative sources, and referred those involved to their regulators.

    FinalFabricated citations
  19. South Korea

    Seoul High Court · Kakao Mobility v. Korea Fair Trade Commission

    Seoul High Court annuls KFTC sanctions on Kakao Mobility over its taxi-dispatch algorithm

    The Seoul High Court quashed a KRW 27.12 billion fine and orders over alleged favouring of Kakao T Blue taxis in AI-assisted dispatch; appeal pending.

    Under appealCompetitionPlatform work
  20. United States

    U.S. District Court for the Middle District of Florida · Garcia v. Character Technologies, Inc.

    Product liability claims against an AI chatbot company survive a First Amendment challenge

    A Florida federal court declined, at the pleading stage, to treat Character.AI’s chatbot output as protected speech and let most claims proceed.

    SettledChatbot liabilityConsumer protection
  21. South Korea

    Personal Information Protection Commission (PIPC) · DeepSeek

    DeepSeek told to delete Korean users' prompts sent abroad and to regularise its overseas transfers

    The PIPC found DeepSeek had sent Korean users' data, prompts included, to firms in China and the US without notice, and told it to delete the prompts.

    FinalData protection
  22. Türkiye

    Turkish Competition Board · Amazon Turkey Perakende Hizmetleri Ltd. Şti. (otomatik fiyatlandırma mekanizması)

    Trendyol and Hepsiburada change their automated repricing tools under commitments, while no infringement is found against Amazon

    In a probe into sellers' repricing algorithms, two platforms gave commitments; Amazon was cleared, partly because its tool was rule-based.

    FinalCompetition
  23. Italy

    Italian Data Protection Authority (Garante per la protezione dei dati personali) · Luka Inc. (Replika)

    EUR 5 million fine on Replika's developer for missing legal basis and inadequate age verification

    The Garante fined Luka Inc., operator of the companion chatbot Replika, EUR 5 million for failings on legal basis, transparency and age verification.

    Under appealData protectionChatbot liabilityConsumer protection
  24. United States

    U.S. Court of Appeals for the D.C. Circuit · Thaler v. Perlmutter

    Copyright Act requires a human author; AI-only image refused registration

    The D.C. Circuit held that the Copyright Act requires every eligible work to be authored in the first instance by a human being.

    FinalAI authorship
  25. European Union

    Court of Justice of the EU (CJEU), First Chamber · CK v Magistrat der Stadt Wien and Dun & Bradstreet Austria GmbH

    People subject to automated credit assessments are entitled to an intelligible explanation

    The CJEU held that controllers must explain intelligibly how personal data shaped an automated decision; trade secrets do not bar access outright.

    FinalAutomated decisionsTransparencyData protection
  26. United States

    U.S. District Court for the District of Delaware · Thomson Reuters Enterprise Centre GmbH v. ROSS Intelligence Inc.

    Training a competing legal search tool on Westlaw headnotes held not to be fair use

    A Delaware federal court rejected ROSS Intelligence’s fair-use defense for using Westlaw headnotes to train a competing AI legal research tool.

    Under appealCopyright and training
  27. Japan

    Intellectual Property High Court of Japan · DABUS (出願却下処分取消請求控訴事件)

    AI-made inventions cannot be patented under Japan’s current Patent Act

    Japan’s IP High Court held that the Patent Act grants patent rights only for inventions made by natural persons, so DABUS cannot be named inventor.

    FinalPatents
  28. South Korea

    Personal Information Protection Commission (PIPC) · Worldcoin Foundation and Tools for Humanity

    Worldcoin Foundation and Tools for Humanity sanctioned over iris data collected in Korea

    The PIPC imposed KRW 1.104 billion on the Worldcoin Foundation and Tools for Humanity for processing iris data without proper consent and sending it abroad.

    FinalBiometricsData protection
  29. United States

    U.S. District Court for the Northern District of California · Mobley v. Workday, Inc.

    AI screening vendor may be liable as an employer’s agent under anti-discrimination law

    The court held that Workday, to which employers allegedly delegated applicant screening, could be liable as their agent under federal anti-discrimination law.

    Interim rulingAutomated decisionsDiscrimination
  30. China

    Beijing Internet Court · Yin v. Beijing technology company et al. (殷某某诉北京某智能科技公司等)

    Voice actor’s rights in their voice extend to an AI-generated voice

    The Beijing Internet Court held that protection of a person’s voice extends to an identifiable AI-generated voice and awarded RMB 250,000.

    FinalPersonality rightsDeepfakes
  31. United States

    Superior Court of Washington for King County · State of Washington v. Puloka

    AI-enhanced video excluded as evidence in a murder trial

    A Washington trial court refused to admit a defense video enhanced with AI because the technique lacked general acceptance in forensic video analysis.

    Proceedings ongoingCriminal law
  32. United States

    U.S. Federal Trade Commission (FTC) / U.S. District Court for the Eastern District of Pennsylvania · FTC v. Rite Aid Corporation

    Retailer banned from using facial recognition for five years

    The FTC secured a five-year ban and extensive safeguards after Rite Aid’s facial recognition system wrongly flagged customers as shoplifters.

    FinalBiometricsDiscriminationConsumer protection
  33. Canada

    British Columbia Civil Resolution Tribunal · Moffatt v. Air Canada

    Airline held liable for inaccurate information given by its chatbot

    The tribunal held Air Canada liable for negligent misrepresentation after its website chatbot gave wrong information about bereavement fares.

    FinalChatbot liabilityConsumer protection
  34. United Kingdom

    UK Supreme Court · Thaler v Comptroller-General of Patents, Designs and Trade Marks

    The AI system DABUS cannot be an inventor under the Patents Act 1977

    The UK Supreme Court unanimously held that an inventor must be a natural person and that owning the machine gives no right to a patent.

    FinalPatentsAI authorship
  35. European Union

    Court of Justice of the EU (CJEU), First Chamber · OQ v Land Hessen (SCHUFA Holding – Scoring)

    A credit score that drives the lender's decision is itself an automated decision

    The CJEU held that a SCHUFA score on which banks rely heavily when deciding on credit is an automated individual decision under Article 22 GDPR.

    FinalAutomated decisionsData protection
  36. China

    Beijing Internet Court · Li v. Liu (李某诉刘某)

    User recognized as author of an image generated with AI

    The Beijing Internet Court held that an image made with Stable Diffusion through the user’s prompts and parameters is a work authored by that user.

    FinalAI authorship
  37. Türkiye

    Advertisement Board (Ministry of Trade) · ChatGPT tarafından oluşturulan üstünlük iddialı tanıtımlar

    Ads carrying superiority claims generated by ChatGPT suspended

    The Advertisement Board suspended ads in three cases that used unsubstantiated superiority claims generated by ChatGPT.

    FinalConsumer protection
  38. South Korea

    Personal Information Protection Commission (PIPC) · OpenAI OpCo, LLC (ChatGPT)

    OpenAI fined for failing to report a ChatGPT data leak affecting Korean users on time

    The PIPC fined OpenAI KRW 3.6 million for not reporting within 24 hours a ChatGPT Plus leak that exposed data of 687 Korean users.

    FinalData protection
  39. Türkiye

    Turkish Competition Board · DSM Grup Danışmanlık İletişim ve Satış Ticaret AŞ (Trendyol)

    Trendyol fined for adjusting its marketplace algorithm to favour its own private-label products

    The Board fined Trendyol TRY 61.3 million for algorithm interventions and use of seller data, and imposed algorithm-related obligations.

    FinalCompetition
  40. Council of Europe (ECHR)

    European Court of Human Rights (ECtHR), Third Section · Glukhin v. Russia

    Using facial recognition to identify and arrest a peaceful solo protester breached Articles 8 and 10

    The ECtHR found that identifying and arresting a solo protester in the Moscow underground with facial recognition breached Articles 8 and 10 ECHR.

    FinalBiometricsData protection
  41. United States

    U.S. District Court for the Southern District of New York · Mata v. Avianca, Inc.

    Lawyers sanctioned for filing fake case citations generated by ChatGPT

    A New York federal judge fined two lawyers and their firm $5,000 for submitting non-existent, ChatGPT-generated opinions and standing by them.

    FinalFabricated citations
  42. Türkiye

    Personal Data Protection Board (KVKK) · Yüz tanıma sistemiyle işe giriş-çıkış kontrolü

    Factory fined TRY 500,000 for using facial recognition to clock staff in and out

    The Board held the employees' blanket consent invalid, found facial recognition disproportionate and ordered the biometric data destroyed.

    FinalData protectionBiometrics
  43. South Korea

    Personal Information Protection Commission (PIPC) · Scatter Lab (Iruda chatbot)

    Chatbot developer sanctioned for building its model on users' private messages

    The PIPC found eight breaches in Scatter Lab's use of some 600,000 users' KakaoTalk chats to build the Iruda chatbot and imposed KRW 103.3 million.

    FinalData protection
  44. Italy

    Tribunal of Bologna, Labour Section (Tribunale di Bologna, Sezione Lavoro) · Filcams CGIL Bologna, Nidil CGIL Bologna, Filt CGIL Bologna v Deliveroo Italia S.r.l.

    Deliveroo's rider booking and ranking system held to discriminate indirectly against riders who went on strike

    The Bologna court held that Deliveroo's booking system, blind to the reasons for absence, put riders who joined strikes at a disadvantage.

    Proceedings ongoingDiscriminationPlatform workAutomated decisions
  45. Netherlands

    District Court of The Hague (Rechtbank Den Haag) · NJCM et al. v State of the Netherlands (SyRI)

    SyRI risk-profiling legislation for welfare and tax fraud held incompatible with Article 8 ECHR

    The Hague District Court declared the SyRI fraud-detection legislation non-binding because the system was insufficiently transparent and verifiable.

    FinalPublic sectorAutomated decisionsTransparency
  46. Türkiye

    Kızılcahamam Civil Court of First Instance · Yapay zekâ ile hazırlanan dilekçede hatalı Yargıtay künyeleri

    Court refers lawyer to the bar and prosecutors over AI-generated incorrect Court of Cassation citations

    When the cited Court of Cassation references could not be found on UYAP, the court reported the lawyer to the bar and prosecutors the same day.

    Proceedings ongoingFabricated citations