AI BriefingAugust 2026 Dear Sir or Madam,The summer of AI in 2026 has brought a wide range of legal developments:The courts have begun to cut a clear path through the legal thicket. The judgment of the Munich I Regional Court against Google offers a ray of hope to anyone who finds themselves misrep
 ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ 
 ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌   ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ 

AI Briefing

August 2026

Dear Sir or Madam,


The summer of AI in 2026 has brought a wide range of legal developments:


The courts have begun to cut a clear path through the legal thicket. The judgment of the Munich I Regional Court against Google offers a ray of hope to anyone who finds themselves misrepresented in an “AI Overview”. The same court also ruled in favour of the rights holders in the second action brought by GEMA against an AI provider. It applied not only German copyright law but also US copyright law. The Munich I Regional Court is thus likely to attract the attention of AI providers even beyond Germany’s borders.


The AI Act is also attracting international attention. The impact of the transparency obligations taking effect under the AI Act are already visible and noticeable across all media. Critics' early warnings now seem to be materialising: the sheer volume of labeling disclosures risks diluting their original warning function.


At least this foreword was written without the assistance of AI. Details of the latest developments can be found below. They may contain traces of AI but have been lovingly curated and checked by hand.


Kind regards,




Dr Andreas Lober

Head of the IP, IT & Data Practice Group

vCard

+++ TRANSPARENCY OBLIGATIONS NOW APPLY +++ MUNICH I REGIONAL COURT: GOOGLE LIABLE FOR FALSE STATEMENTS IN AI OVERVIEWS +++ MUNICH I REGIONAL COURT: GEMA PREVAILS AGAINST AI MUSIC GENERATOR SUNO +++

1. Changes in Legislation

+++ GERMAN AI MARKET SURVEILLANCE AND INNOVATION PROMOTION ACT ENTERS INTO FORCE AND ESTABLISHES SUPERVISORY STRUCTURE UNDER THE AI ACT +++


The German legislation implementing the AI Act entered into force on 29 July 2026 as the AI Market Surveillance and Innovation Promotion Act (KI-MIG). As a general rule, it designates the Federal Network Agency as the competent market surveillance authority, single point of contact and complaints body, unless specific responsibilities apply. In particular, the existing product surveillance authorities remain responsible for AI systems connected to harmonised products. The Act also assigns responsibility to the Federal Financial Supervisory Authority for certain AI systems connected with regulated financial activities. In addition, the KI-MIG governs cooperation between authorities, their powers, AI regulatory sandboxes, the real-world testing of high-risk AI systems, fines and record-retention requirements. Germany’s implementation and supervisory structure under the AI Act is now in place (see also AI Briefing November 2025).


To the Act in the Federal Law Gazette (28 July 2026, Federal Law Gazette 2026 I No. 223, in German)


To the Federal Network Agency press release (29 July 2026, in German)


+++ EU: DIGITAL OMNIBUS AMENDING THE AI ACT ENTERS INTO FORCE +++


The amendments to the AI Act under the Digital Omnibus entered into force on 27 July 2026. Regulation (EU) 2026/1744, in particular, postpones the dates from which the requirements for certain high-risk AI systems will apply. The new date for high-risk use cases listed in Annex III to the AI Act is generally 2 December 2027. For high-risk AI systems that are safety components of certain products, or are themselves such products, the new date is 2 August 2028. In addition, certain simplifications for small and medium-sized enterprises are extended to small mid-cap companies, AI literacy obligations are simplified, further testing and experimentation opportunities are created, and the powers of the AI Office are expanded. The legislative process has thus been completed (see AI Briefings February and May 2026).


To Regulation (EU) 2026/1744 (8 July 2026)


To the European Commission communication (27 July 2026)

 

2. Case law

+++ MUNICH I REGIONAL COURT: GOOGLE LIABLE FOR FALSE STATEMENTS IN AI OVERVIEWS +++


The Munich I Regional Court has ordered Google to cease disseminating factual claims about two publishing houses through its “AI Overview” feature. Unlike conventional search results, Google did not simply find and display third-party content. Instead, the AI Overview independently summarised and restructured information, and used it to formulate an answer to the search query. The answer was thus to be considered as content produced by Google itself. To that extent, the provider could not rely on liability privileges for third-party content.


To the Munich I Regional Court press release on the judgment (28 May 2026, 26 O 869/26, in German)


+++ FRANKFURT REGIONAL COURT: SIMILAR AI-EDITED PRODUCT IMAGE DOES NOT AUTOMATICALLY INFRINGE COPYRIGHT +++


The Frankfurt Regional Court has dismissed an action concerning similarly designed product images of cable glands. The claimant had assumed that the defendant had entered the claimant’s product photograph into an AI system and generated its own image on that basis. However, the fact that test runs using the claimant’s photograph produced a similar result was not sufficient to prove that the defendant had in fact used that photograph as input. The defendant had submitted its own source photographs. Even assuming that the claimant’s photograph had been used, the Court found no interference with the narrow scope of protection afforded to the product photograph as a simple photograph. Uploading the photograph to the AI system could potentially amount to an unauthorised reproduction but this was not at issue in the proceedings. The Court also held that another image, specifically a computer-generated CAD rendering, was neither protected as a photograph nor, in the absence of sufficient originality, as a copyright work.


To the judgment of Frankfurt Regional Court (27 May 2026, 2-06 O 347/25, in German)


+++ HIGHER ADMINISTRATIVE COURT OF NORTH RHINE-WESTPHALIA: POTENTIAL FUTURE USE OF AI TO ENHANCE AERIAL IMAGES DOES NOT AUTOMATICALLY LEAD TO A GREATER INFRINGEMENT OF FUNDAMENTAL RIGHTS +++


The Higher Administrative Court of North Rhine-Westphalia has dismissed an appeal concerning aerial images taken by a municipality to assess stormwater charges. The claimant argued, among other things, that the relatively blurred images of its property could in future be processed and sharpened using AI, and combined with other data. In the Court’s view, the mere technical possibility of such processing did not automatically mean that taking and using the aerial images already constituted an infringement of fundamental rights of more than a minor nature. There was no specific indication that the municipality or its service provider intended to carry out AI-assisted post-processing. Nor did the abstract possibility of data theft and subsequent processing by third parties alter that assessment.


To the decision of the Higher Administrative Court of North Rhine-Westphalia (22 June 2026, 16 B 169/25, in German)


+++ MUNICH REGIONAL COURT I: GEMA PREVAILS AGAINST AI MUSIC GENERATOR SUNO +++


Munich I Regional Court largely upheld GEMA’s claims for injunctive relief, information and damages against the provider of the Suno AI music generator. The case concerned six protected musical works that had undisputedly been used to train the model. In the Court’s view, the works had not only been used during the training process. Rather, their distinctive musical elements had been encoded in the model parameters in such a way that they could be reproducibly retrieved through suitable inputs. The Court regarded this “memorisation” as an embodiment of the works in the model and therefore as a reproduction relevant under copyright law. It also found further reproductions in the challenged outputs, which were sufficiently similar to the original works. The storage of the works in the model was not covered by the text and data mining exception under section 44b of the German Copyright Act. The Court applied US copyright law to the training activities carried out in the United States and rejected the fair use defence.


To the Munich I Regional Court press release (31 July 2026, 42 O 763/25, in German)


+++ GENERAL COURT: “OPENAI” CANNOT BE REGISTERED AS AN EU TRADE MARK FOR AI-RELATED GOODS AND SERVICES +++


The General Court of the EU has dismissed OpenAI’s action against the partial rejection of its application to register “OPENAI” as an EU trade mark. The relevant English-speaking public would understand the sign, consisting of the elements “open” and “AI”, as referring to freely accessible or unrestricted artificial intelligence. In relation to the software and IT services in question, the sign therefore directly described a characteristic or intended use of the goods and services. Combining the two common elements without a space or hyphen does not convey a meaning that goes beyond the sum of its parts. Consequently, the sign is descriptive and lacks sufficient distinctive character for the goods and services in question.


To the judgment of the General Court (15 July 2026, T-555/25)

 

3. Opinions / Other

+++ EUROPEAN COMMISSION: ARTICLE 50 AI ACT TRANSPARENCY OBLIGATIONS NOW APPLY; FINAL GUIDELINES AND CODE OF PRACTICE PUBLISHED +++


The transparency obligations under Article 50 of the AI Act have applied since 2 August 2026. Providers of certain interactive AI systems must ensure that individuals are informed that they are interacting with an AI system. Providers of generative AI systems must also ensure that their outputs are marked in a machine-readable format as artificially generated or manipulated. Deployers are subject, among other things, to information obligations when using emotion recognition and biometric categorisation systems, as well as disclosure obligations for deepfakes and certain AI-generated text on matters of public interest. On 20 July 2026, the European Commission published final guidelines on these requirements. A voluntary Code of Practice supplements the guidelines by specifying technical and visible measures for labelling AI-generated content (see AI Briefings February and May 2026).


To the European Commission guidelines (20 July 2026)


To the Code of Practice on AI-generated content


To the European Commission communication on the start of enforcement (31 July 2026)

Beiten Burkhardt Rechtsanwaltsgesellschaft mbH is a member of ADVANT, an association of independent law firms. Each Member Firm is a separate and legally distinct entity, and is liable only for its own acts or omissions.

EDITOR IN CHARGE
Dr Andreas Lober | Rechtsanwalt
©Beiten Burkhardt
Rechtsanwaltsgesellschaft mbH
www.advant-beiten.com

Your contacts of the Team

 

Please note: If you no longer wish to receive information, you can unsubscribe at any time.

Imprint

Beiten Burkhardt Rechtsanwaltsgesellschaft mbH
Ganghoferstraße 33, 80339 München
Registered unter HR B 155350 at the Regional Court Munich / VAT Reg. No. DE-811218811

Tel.: +49 89 35065-0, Fax: +49 89 35065-123 | E-Mail: munich@advant-beiten.com
Here you will find our complete imprint: www.advant-beiten.com/imprint 
and our privacy policy: www.advant-beiten.com/en/privacy-protection

View in Browser