Innovation Law Insights

Podcast

Daniela Paletti of Condé Nast on the future of media and its legal challenges

In this compelling episode of the podcast Diritto al Digitale, hosts Giulio Coraggio and Elena Varese from DLA Piper sit down with Daniela Paletti, the Head of Legal Europe at Condé Nast. Watch it here.

 

Artificial Intelligence

AI advances in the US: Navigating innovation roadmaps and new legislation

A few months following the adoption of the AI Act in the EU, a bipartisan group of UA senators has introduced a new roadmap to guide Congress in regulating and advancing AI and promoting innovation.

In parallel, almost all US states, most recently Colorado, are trying to introduce AI-related bills, many of which address deepfake issues, although only a few have passed into law.

Innovation and legislative priorities

The 20-page report, “Driving US Innovation in Artificial Intelligence” was released on May 15 by the US senate AI working group. It focuses on promoting innovation while addressing potential risks associated with AI technology. Unlike the AI Act adopted by European neighbours, this roadmap aims to inform legislative efforts without imposing strict regulations that, from a US perspective, could stifle growth.

The roadmap recommends that Congress allocate USD32 billion annually for AI innovation, a figure proposed by the National Security Commission on Artificial Intelligence. The importance of this funding was emphasized to maintain the US’s leadership in AI, particularly in sectors like healthcare in which AI’s revolutionizing potential is only beginning.

At the same time, the roadmap recommends increasing regulatory efforts. As the culmination of a series of “AI insight forums” that included over 150 experts from various industries, the roadmap doesn’t contain specific legislative proposals but aims to serve as guidance for lawmakers.

Key areas of focus where targeted legislation have been encouraged include:

• Transparency requirements for AI systems that balance innovation with risk management.
• Incentives for developers to provide content provenance information for AI-integrated software.
• Protections against the unauthorized use of an individual’s name, likeness, image or voice.
• Increasing public awareness of AI’s role in daily life.
• Research and development tools to help manage AI risks, particularly in high-risk sectors like finance, healthcare, and housing.

The document emphasizes the need for federal privacy legislation to address AI’s reliance on personal data. It supports the American Privacy Rights Act, which addresses data minimization, security, consumer rights, data brokers, and consent requirements. The working group suggests reviewing existing laws to determine their adequacy in addressing AI-related issues, especially given the opaque nature of some AI systems.

Comprehensive approach and AI-issues

The working group has urged Congress to collaborate on AI legislation, establish clear definitions for key terms, and stay informed about executive branch AI initiatives. It calls for cross-government AI research, infrastructure support for the National Institute of Standards and Technology, and enhanced capabilities for the Bureau of Industry and Security.

The roadmap also addresses a number of concerns emerging as a consequence of fast AI developments, particularly:

• Protection for children: the group recommended developing legislation to protect children from potential AI-powered harms online by ensuring companies take reasonable steps to consider such risks in product design and operation. The AI Working Group is concerned by data demonstrating the mental health impact of social media and expresses support for further study and action by the relevant agencies to understand and combat this issue.
• Social scoring: the group emphasized the risks of AI for social scoring and encouraged developing legislation to ban this use and protect US citizens’ fundamental freedom.
• Workforce concerns: the group emphasized the need to train and upskill employees to adapt to an AI-driven economy. They recommend leveraging existing programs like the US Digital Service and the Presidential Innovation Fellows to attract and retain AI talent.
• Deepfake regulation: the Senate Rules Committee recently advanced bipartisan legislation to safeguard elections from AI-generated deceptive content, including deepfakes. The roadmap supports these efforts and calls for regulations to protect against the non-consensual use of AI-generated intimate images.

Finally, in light of the varying risks that AI systems may present, the working group urges thorough testing of AI systems to prevent harms and advises against releasing those that don't meet standards. They recommend flexible risk assessment methods focusing on AI capabilities, protecting proprietary info, and fostering US AI innovation.

Emerging national AI legislation

While no major AI legislation has yet passed Congress, the roadmap lays the groundwork for future bipartisan efforts to regulate and harness AI technology.

In the absence of clear rules at federal level, US states are swiftly introducing their own AI bills. In March 2024, Utah was the first state to enact an AI-focused statute governing tackling the use of generative AI for consumer deception. Other states – including California, Connecticut, Delaware and Indiana – have resorted to their comprehensive data privacy laws to include provisions that in some way regulates the use of AI in profiling. On 8 May Colorado state lawmakers passed a landmark AI bill that shares notable common ground with the EU AI Act, by adopting a risk-oriented approach towards AI, setting regulations for high-risk systems, and outlining guidelines for disclosing the use of AI.

Numerous other proposals are being discussed at various national levels. But the primary concern is the possibility of a fragmented policy landscape in the absence of a federal framework. Fragmentation, rather than fostering innovation, could lead to a convoluted and intricate regulatory system.

Author: Maria Chiara Meneghetti

Free and Open Source Software and AI: A pairing for open innovation

In the constantly evolving digital era, the intersection between Free and Open Source Software and AI emerges as a crucial focus of technological innovation. Our analysis aims to explore the dynamic and fruitful link between these two driving forces of the digital process through an investigation into the philosophical foundations of Free Software and its entanglement with advances in AI.

Free and Open Source Software (FOSS)

FOSS is an alternative to the closed software model associated with most commercial software license. The FOSS model promotes innovation and free circulation of source code, as opposed to the proprietary software model which is based on the owner’s prerogative to prohibit third parties from exploiting its rights.

Free Software is essentially based on the idea that software should be free to use and non-free software is a social problem and an obstacle to innovation. According to the approach of the Free Software Foundation, a piece of software is “free” if it respects the fundamental freedoms of users and community:

• the freedom to run the program at will and for any purpose;
• the freedom to study how the program works and to modify it to suit specific needs;
• the freedom to redistribute copies;
• the freedom to make improvements to the program and to distribute them publicly, so that the whole community can benefit.

Open Source software has a more pragmatic connotation, focusing on the accessibility of the source code. The distribution of open source software must be free, the license must allow the rights to sell or donate it, and it must not provide royalties for its sale. Furthermore, the license must allow the modification of the original programs and the creation of derivative programs as well as the distribution of these under the terms of the original software license. Moreover, the license may only provide for restrictions on distribution of modified source code if patch files are distributed with the original code, and it must not discriminate against persons or groups or areas of use. The rights to the program must apply to all those to whom it is distributed, without need to issue additional licenses. If the licensed program is part of a software distribution, the license must not be specific to that, but it must guarantee to any recipient the same rights of the original distribution. The license must also not contaminate other software distributed with the main one and it must respect the principle of technological neutrality.

Despite the differences between the two, it’s common in practice to use the terms “free software” and “open source software” interchangeably and it is not uncommon to refer to this phenomenon using the term Free and Open Source Software.

FOSS licenses

More than 40 licenses coexist in the current landscape, differing in both the rights granted and the conditions imposed. FOSS licenses can be categorized according to the different level of protection granted to the software:

• Restrictive licenses (also known as copyleft licenses) place restrictions on the redistribution of derivative works to ensure the opening of code. They grant users of the software the right to use, copy, modify, improve and redistribute as long as conditions of reciprocity are guaranteed: in other words, they prevent pejorative conditions from being applied, to prohibit the emergence of proprietary model. An example is the GPL license.
• Permissive licenses (also known as non-copyleft licenses) allow broad use of source code even in non-open source programs (including possible modification) and they don’t restrict the distribution of derivative works. BSD and Apache are two examples of non-copyleft licenses.

Between the two extremes lie certain grey-area licenses with weaker copyleft clauses, such as the LGPL or MPL.

By far the most popular open source licenses for AI-based software releases are permissive licenses.

Advantages of the FOSS-AI intersection

The entanglement of the worlds of FOSS and AI is a turning point for the ongoing technological revolution. The FOSS community fosters the development of increasingly advanced AI systems, allowing public or private actors to have rapid access to resources, fostering the innovation process, without the need for huge initial investments associated with proprietary software licenses.

Examining the source code provides insight into the working mechanisms of AI systems, increasing transparency and levels of user confidence: if the algorithms are understandable and verified, a high degree of fairness and constant prevention of bias may be guaranteed.

Models and code can be customized according to specific requirements, offering a flexibility that simple open source solutions in other contexts can’t provide.

The intersection of AI and FOSS also promotes the creation of a collaborative ecosystem in which developers can share ideas, contribute to improvements and accelerate technological progress.

Finally, the availability of the source code increases the level of security, allowing constant monitoring and subsequent revision of systems functional to a rapid response to threats in cyberspace.

Implementing open source AI

An AI-based system consists of more components than traditional software and the definition of open source needs to adapt and expand: in the context of AI, there’s no source code per se, and the key element is the data used to train the system.

In the case of proprietary AI systems, a “per token” charge is usually collected, and charging levels vary depending on the models and their responsiveness. Charges often include hosting and support costs and charge per token is highly susceptible to upward fluctuations. Open source AI systems are often downloadable free of charge and despite their expensiveness at least in the installation phase, they prove to be cheaper for customers who make heavy use of them.

There are many possible uses of an open source AI and the most common scenarios include calling via APIs or function calls, incorporation, linking, modification and translation. It’s very common for software to be both proprietary and open source, and in most cases API calls (ie processes allowing two pieces of software to exchange data) or function calls are used. In such cases, no contamination or mixing of source code is necessary. A developer could also copy part of a FOSS within a proprietary product: here the incorporation occurs. This results in contamination. And in copyleft licenses all source code is required to be open if the resulting product is to be distributed. A developer might often link a FOSS component with a proprietary product: in this case, there’s no contamination of the source code. Similar to API calls, linking to AI models from other source code is a very common form of implementation. Open code allows developers to make changes to a FOSS component, adding new FOSS components, correcting and optimizing the original components and deleting certain parts of code. Finally, source code might be translated, for instance from one programming language to another.

AI Act and Product Liability Directive: The future of the European approach to FOSS

In its latest approved version, the European Artificial Intelligence Regulation (AI Act) has led to confusion as to the application to open source software. Initially, Art. 2 (5)(g) of the 26 January 2024 version excluded the Regulation’s obligations for AI systems released under FOSS licenses, unless they were high-risk or fell under Titles II and IV. But the new version regulated the same issue in Art. 2 para. 12, making the Regulation applicable to open source systems. An interpretation error in the last revision led to the belief that FOSS might not be exempted, but a subsequent clarification indicated that corrections will be made to correct this error in the final version of the text published in the Official Gazette of the European Union. The free software will benefit from the exemption if its parameters (including information on model architecture) are made public and if it’s not made available against payment or anyway monetized (eg through the paid offer of technical support following the release). To get a clear idea behind the European approach, it’s worth looking at Recitals 102 and 103 of the Regulation’s latest version: they state that an AI system released under a FOSS license allowing it to be shared openly and users to freely access, use, modify and redistribute data or modified version, should not be covered by the Regulation, as it contributes to market research and innovations offering significant growth opportunities for the EU economy.

The new Product Liability Directive, which is part of the package of EU measures aimed at supporting the promotion of AI, also excludes FOSS from its application. This choice, once again, appears justified the encouragement of research and innovation in the European market. Finally, free software, not developed or supplied in the course of a commercial activity, is by definition not “placed on the market.” Where the software is provided for remuneration or personal data is used in the context of a commercial activity, the Directive is applicable.

Author: Alessandra Faranda

 

Food and Beverages

ICQRF issues report on the activities carried out in 2023

The Central Inspectorate for Quality Protection and Fraud Repression of Agri-Food Products (ICQRF) recently made available the Report concerning the activities performed in 2023.

Analysing the document is always particularly useful to the food and agriculture operators, as it provides an overview of the Inspectorate's areas of intervention.

The ICQRF is the Italian authority responsible for the protection of Protected Designations of Origin (PDOs) and Protected Geographical Indications (PGIs). It also imposes administrative fines in agricultural and agribusiness matters under state jurisdiction. In the report, it describes its main activities for 2023.

Once again, the numbers related to the ICQRF’s actions are particularly interesting. The report shows that the ICQRF carried out more than 54,000 anti-fraud inspections and issued as many as 5,548 administrative charges, with a total value of goods seized of more than EUR42million.

The ICQRF is also the national enforcement authority in charge of investigating violations of the implementing legislation of the Unfair Trade Practices (UTP) Directive, the Legislative Decree 198/2021, which came into force to protect suppliers of agricultural and food products in their commercial relationships with buyers.

In the context of combating unfair trade practices in the agribusiness sector, in 2023 the ICQRF raised as many as 53 administrative challenges in the dairy sector and 15 administrative challenges in the fruit and vegetable sector. The number of administrative challenges in the cereals, wine and oil sectors was slightly lower.

In total, the ICQRF conducted as many as 488 inspections directed at 315 different operators in the sector, finding that as many as 32 economic operators had committed breaches of the Legislative Decree 198/2021. As stated in the Report, most of the challenges raised concerned the failure of buyers of agricultural and food products to comply with payment terms (40 breaches detected), the absence of a written contract concluded before the delivery of the products (9 breaches sanctioned) or the lack of essential contractual elements such as the duration, quantity and characteristics of the product.

The report ends with brief considerations with respect to the role of the ICQRF as the competent Authority for controls related to agri-food commodities (cattle, cocoa, coffee, oil palm, soybeans) in accordance with EU Regulation 2023/1115, concerning the commodities and products associated with deforestation and forest degradation, which came into force in June 2023.

With respect to this area, the European Regulation 2023/1115 provides for the establishment of an electronic interface system to exchange essential information between the competent authority and the customs authority.

Author: Federico Maria Di Vizio

 

Intellectual Property

Trademarks and gin: EU General Court confirms the invalidity of a trademark for descriptiveness of the geographical origin and quality of the designated goods

On 21 February 2024, the General Court of the European Union ruled on an appeal requesting the annulment of the decision of the Fifth Board of Appeal of the European Union Office for Intellectual Property, dated 23 September 2022, issued in case R 1978/2021-5. At the heart of the dispute was the issue of the declaration of invalidity of a trademark registration deemed descriptive of the geographical origin and quality of the goods designated in the application.

The facts: Application for invalidity and appeal before the EUIPO

The dispute involved a well-known Scottish company producing alcoholic beverages, including gin (the defendant) and a Peruvian distillery (the applicant).

On 18 September 2020, the Respondent filed an application with the European Union Intellectual Property Office (EUIPO) for a declaration of invalidity of the European Union trademark “Amazonian GIN COMPANY" (den.) (the contested trademark), filed on 11 September 2018 by the Applicant and granted on 9 January 2019, in class 33 of the Nice Classification, for “ gin (brandy); alcoholic beverages made from gin."

The grounds relied on in support of the application for a declaration of invalidity were those set out in Article 59(1)(a), read in conjunction with Article 7(1)(a)(c), of Regulation (EU) 2017/1001 on the European Union Trade Mark (EUTMR), in the light of which “an EU trade mark shall be declared invalid where it has been registered contrary to the provisions of Article 7', which states that “ (...) shall be excluded from registration where the trade mark consists exclusively of signs or indications which may serve, in trade, to designate the kind, quality, quantity, intended purpose, value, geographical origin, or the time of production of the goods or of rendering of the service, or other characteristics of the goods or services.”

The application for a declaration of invalidity was then rejected on 28 October 2021 by the Cancellation Division. The respondent's subsequent appeal before the EUIPO against that decision was then upheld by the Fifth Board of Appeal (the board) in a decision dated 23 September 2022.

The board upheld the appeal, noting the descriptive nature of the contested mark and, consequently, its infringement of the rule. Briefly, the board found, in essence, that the English-speaking part of the relevant public immediately and exclusively understood the contested trademark as an indication that the registered alcoholic beverages or their ingredients are related to the region around the Amazon River.

The judgment of the General Court of the European Union

The descriptive nature of the contested trademark, for “gin (spirit); alcoholic beverages based on gin” in class 33, was then confirmed by the General Court of the European Union (General Court).

The General Court concluded that the term “Amazonian” is in fact capable of describing not only the geographical origin of the gins, but also the quality of the goods designated. Consequently, it dismissed the applicant's action, confirming the invalidity of the contested trademark.

Regarding the issue of descriptiveness of the geographical origin and quality of the products, the Court emphasized that:

• although the Amazon region extends over several countries, it’s a well-defined geographical area, renowned for its countless botanical species;
• a wide variety of botanicals may be involved in the production of gin; substances that give gin a characteristic aroma and taste that may determine the consumer's choice. Therefore, the consumer will establish a link between the Amazon region and gin;
• the term “Amazonian” can convey a positive image of the botanicals used to produce or flavour gin and, consequently, of its essential characteristics such as taste;
• the gin marketed under the contested trademark is described by the applicant as containing many botanicals and spices from or distilled in the Amazon region. So, although this marketing strategy can’t make a trademark descriptive, it may confirm the findings regarding its descriptive character.

In conclusion, the contested trademark appears to be capable of conveying to the final consumer information regarding the geographical origin and quality of the designated goods and their ingredients and, consequently, in breach of Article 7 EUTMR.

Author: Tamara D'Angeli

---

Innovation Law Insights is compiled by the professionals at the law firm DLA Piper under the coordination of Arianna Angilletta, Matteo Antonelli, Edoardo Bardelli, Carolina Battistella, Carlotta Busani, Giorgia Carneri, Maria Rita Cormaci, Camila Crisci, Cristina Criscuoli, Tamara D’Angeli, Chiara D’Onofrio, Federico Maria Di Vizio, Enila Elezi, Alessandra Faranda, Nadia Feola, Laura Gastaldi, Vincenzo Giuffré, Nicola Landolfi, Giacomo Lusardi, Valentina Mazza, Lara Mastrangelo, Maria Chiara Meneghetti, Deborah Paracchini, Maria Vittoria Pessina, Tommaso Ricci, Miriam Romeo, Rebecca Rossi, Roxana Smeria, Massimiliano Tiberio, Giulia Zappaterra.

Articles concerning Telecommunications are curated by Massimo D’Andrea, Flaminia Perna e Matilde Losa.

For further information on the topics covered, please contact the partners Giulio Coraggio, Marco de Morpurgo, Gualtiero Dragotti, Alessandro Ferrari, Roberto Valenti, Elena Varese, Alessandro Boso Caretta, Ginevra Righini.

If you no longer wish to receive Innovation Law Insights or would like to subscribe, please email Silvia Molignani.