[GuestPost] The role of big data, SEPs, PAEs and copyright in Digital Transformation: IP Tech Summit 2020 Report

Last year, Premier Cercle hosted their IP event in Sweden at the headquarters of Volvo Cars focusing on IP in the automotive industry (see IPKat reports here and here). The year before focused on IP in aerospace (see here). This year’s Summit was hosted by Microsoft and focused more broadly on global digitalization and its impact on IP, highlighting topics such as artificial intelligence, the internet of things, open data and open innovation. Speakers included representatives from large companies such as Nokia, Google and Microsoft as well as policy-makers such as the European Commission and the US Federal Trade Commission. The AmeriKat's friends at Dutch IP firm Brinkhof  - Alexander de Leeuw, Boukje van der Maazen, Jonathan Santman and Jasmijn de Groot - were on hand to report on this year's proceedings. 

Over to the team for their report:

"Although all participants were very much looking forward to visiting Microsoft Germany’s headquarters in Munich, the event had to be held digitally due to the COVID-19 pandemic. The sessions were thought-provoking, inspiring and interactive nonetheless, and provided detailed insights into how some of the world’s leading tech companies are coping with increased digitalization in terms of their IP strategies. In this report we summarize what we consider to be the most notable contributions. 

The important role of data in the digitized world

As can be expected from an IP conference at one of the biggest software companies in the world, one of the overarching topics was data. Data is getting increasingly more attention among IP-practitioners, and rightfully so. In the modern world huge amounts of data are created, processed and stored every day. To put things into perspective, one of the speakers indicated that about 5 million books in data are generated per person each day. This amount increases exponentially over time: the total amount of data generated to date doubles every two years. Who controls this data is a vital issue, as data has value. Accordingly, one of the recurring themes was the so-called “data divide” or “opportunity gap”. According to one speaker, fewer than 100 companies collect 50% of the data, which means that there is an economic power flow to only small number of companies (and countries: the US and China). Microsoft – one of the large fish in the data sea – strives for more data equality with a campaign aimed at closing the data divide (see here).

The development of AI systems also very much depends on data. The key to its performance is the amount and quality of data available for training (such as images, speech, and text). An interesting insight into the usefulness of data was provided by Sahar Attaripour (co-founder of Suguzia AI). She explained that her company uses an AI engine to develop ideas into marketable products. The AI system was described as “Google on steroids”, the trained algorithm being able to search incredible amounts of data (think of all textbooks, central bank reports, research papers, basically anything that can be labelled as knowledge). The algorithm is not only able to search the data, it is capable of understanding its content, making it possible to utilize AI as a tool for the development of ideas to more concrete applications. One of the key difficulties is the storage, organization and consistency of data.

Many of the AI debates concern the protection and licensing of data sets. Access to data can be restricted and can vary from jurisdiction to jurisdiction. Dr. Julia Keim (Microsoft) confirmed that an interesting derivative market has developed where companies claim license fees for the use of their data. The Directive on copyright in the Digital Single Market (2019/790), on the other hand, provides for an exclusion for copyright infringement for data-mining. Medha Rolvering (Global Head of IP at Software AP) expects AI developers to be able to profit from this data-mining exception. 

The future of AI systems depends on successful data-collaboration and complex licensing structures in order to create a navigable landscape. This has (already long ago) sparked the development of open source initiatives. On the one hand, open collaboration was welcomed as a way of stimulating innovation, for example by Keith Bergelt (CEO at Open Innovation Network) who described this as an all-inclusive global phenomenon that is enhancing our ability to create innovation. Stefanie Meliert (Counsel at Fraunhaufer-Gesellschaft), on the other hand, argued that sustainable innovation can only exist when companies focus on value creation and value capture (referring to work by Henry W. Chesbrough). “Innovation for free (in the form of royalty free licenses) is a myth and will only lead to a reallocation of costs and rewards.” What appears to be open source up front, is often compensated for through the back door by paid service models. Open source also does not mean ‘no obligations’. You must carefully track the components you are using to make sure that you continuously comply with the license obligations.

These developments create countless opportunities, but also bring about ethical concerns. There are no clear and strict ethical guidelines in place yet, but a legal framework is beginning to take shape, for example with recently published guidance from the European Commission (see here and here).

Patenting Artificial Intelligence (AI) & Computer Implemented Inventions (CII)

The IP Tech Summit had an impressive amount of experts on AI and CII patenting. Before going into what they had to say about IP, one cannot help but wonder what we mean when we mention the buzzword “AI”.

According to Thierry Paul (Director IP at Element AI) AI is anything (or rather: any task) that can be accomplished by a trained model based on machine learning and deep learning (the latter being the distinguishing element over ‘traditional’ software). Going one step further, Prof. Michael Bronstein (Chair in Machine Learning and Pattern Recognition at Imperial College) explained that our understanding of artificial intelligence is linked to human intelligence: visual perception, language understanding and creativity. AI distinguishes itself over previous technology in terms of its complexity and its capabilities. 

Does this require changes in the law? Opinions were strongly divided on whether existing legal systems have adequate flexibility to accommodate the different aspects related to the development of AI (such as gathering and utilizing training data, developing the algorithm/model, and utilizing the output). Aurélie Jiminez (Member of EPO Board of Appeal 3.1.0.1) is of the opinion that the existing legal framework (specifically talking about the EPC) is flexible enough and that we are still talking about CII, regarding which extensive case law has already been developed. Others argued that current patent systems are too slow for fast-paced technology such as AI (also described as a “moving target”), making patenting infeasible at least for start-up companies. Martin Müller (President of EPO Board of Appeal 3.5.06) added that we may have to reconsider the statutory exclusions from patentability if AI inventions are excluded solely because the contribution lies within mathematics. 

It quickly became apparent that one’s ability to navigate through existing patent law in order to protect an AI/CII invention depends greatly on claim drafting, both in Europe and in the US. The fact that there is no standardized terminology, makes an already difficult topic even more complex. In any event, it was recommended to avoid mathematical equations and terms like ‘neural networks’. Another difficulty is that AI systems largely depend on data, which is difficult to include in a patent application, and that it can be problematic to reproduce an AI achievement due to the large amount of parameters involved (the so-called “replication crisis”). This requires guidance from patent offices. The EPO has already issued guidance on this topic (see here), and guidelines will soon follow from the USPTO (see here). Likewise, Dr. Ulrike Till (Director of the AI Policy Division at WIPO) hinted at the near arrival of guidelines, strategies and instruments applicable to AI and IP prepared by WIPO. 

Finally, what would an AI session be without a debate on DABUS, the first named AI-inventor on a patent application (see here)? Although this was welcomed as an interesting attempt to develop the rules on AI and patents, most panelists agreed that the EPO in the currently pending appeal will not allow an AI-system to be named as the inventor on a patent application. This point of view was also recently adopted in AIPPI Resolution Q272 (see here), which states: “An AI should not be considered an inventor or a co-inventor of an invention, nor be permitted to be named as such, even if no contribution to the invention by a natural person is identifiable.” The question of AI-inventorship was described as more important for the US, where a patent can be deemed unenforceable if it does not identify the inventor correctly.

Standard Essential Patents (SEPs)

With a steep increase in standard essential patents over the past decade, this was a hotly debated topic at this year’s conference. 

Thierry Breton (EU Commissioner for the Internal Market) stressed in his keynote speech that “the licensing of standard essential patents is a cumbersome and costly exercise for both patent holders and technology implementers” and that “there is a need for a much clearer and much more predictable framework incentivizing good faith negotiations rather than recourse to litigation.” According to Thierry, this is essential in order to ensure the EU’s lead position in standardization and the internet of things. He referred to the recently adopted IP action plan (available here), which includes initiatives to improve the framework regarding essentiality declarations, licensing and enforcement of SEPs, including a pilot study for the essentiality assessment of SEPs (see here), and further indicated that the European Commission will enter into a sectorial dialogue with industry players in order to find appropriate solutions. 

Jenni Lukander (President at Nokia Technologies) welcomed the European Commission’s IP action plan, but indicated that industry itself can do better with trying to come up with solutions to the issues surrounding SEP licensing. Eeva Hakoranta (Chief Licensing Officer at Interdigital) also believes that the Commission’s IP action plan is a step in the right direction. At the same time she emphasized the difficulties associated with hold-out behavior, and pleaded for sanctioning this behavior: “a company that refuses over a prolonged period of time to follow the rules of FRAND should lose their right to FRAND and either have to pay supra-FRAND royalties or lose the right to operate in that market”. 

Jenni Lukander also touched upon the recent referral to the CJEU by the Düsseldorf court about component level licensing (see here) when discussing the Nokia/Daimler case. She explained that Nokia generally offers licenses both directly to the car manufacturer and to the tier one suppliers, because some car manufacturers indicated a preference for the latter. Nokia further makes its portfolio available for licensing through the Avanci patent pool. These options are also available to Daimler and its suppliers. She therefore noted that the recent questions referred to the CJEU “while interesting academic questions”, will likely not solve Nokia’s commercial dispute with Daimler. Regarding the CJEU referral she expressed the view that “from a competition point of view, it would be a healthier outcome for the industry if the end product manufacturer who gets the most profit from using these technologies at least took part in sharing the fair royalty cost instead of forcing that cost burden solely to its supplier”. This because, “those who resist licensing the most are some of the most powerful companies in the world and use their market power against their suppliers.”

Regardless of who should take the license, the recurring question and main source of dispute is determining what is a FRAND rate. This is especially the case when there is a lack of transparency regarding the calculation of rates. Recognizing that the comparable licenses approach has it difficulties because licenses are typically concluded on a non-disclosure basis, Kent Baker (Head of IP Strategy and Licensing at U-Blox, a component manufacturer for inter alia the automotive industry) suggested to base the calculation on established economic principles. There was also criticism on transparency as a “fix-all” solution, as according to Eeva Hakoranta (Interdigital) increased transparency has not yet resulted in a difference in willingness of implementers.

Ludwig von Reiche (IP2Innovate, an industry association) talked about problems with the enforcement of SEPs from an implementer point of view. NPEs that are only interested in financial gain are getting increased leverage over implementers as they acquire more and more patents. For example, a patent relating to a 25 dollar chip can be used to obtain an injunction against an entire car. The granting of automatic injunctions in countries like Germany aggravates the problem. IP2Innovate therefore pleads on behalf of its members (companies from a range of industry sectors, such as BMW, Intel and Microsoft), for a genuine proportionality assessment to be applied consistently across European jurisdictions. The German patent reform is seen as a positive development, but according to IP2Innovate more should be done (see also here). 

Even if some balance is found in terms of injunctions, other problems are lurking around the corner for SEPs, such as the increase in anti(anti)suit injunctions and the fact that they create an unwanted race to court. Arbitration was generally viewed as a potential solution, but no concrete proposals were formulated by the speakers. Even if these jurisdictional issues could be solved, another issue facing the telecoms industry is a more political one: restrictive trade policies such as those of the US and China. The need to address these issued continues to increase with the launch of 5G networks. In short: enough topics for more discussions about SEPs!

Patent Assertion Entities (PAEs) 

Are they the uninvited guest to the IP party? Everybody talks about them, but not a lot of conferences actually engage with them. Companies who are primarily in the business of patents to monetize them through licenses: PAEs (also known as Non Practicing Entities, NPEs). One of the sessions at the IP Tech Summit discussed the role of these companies in shaping the future of IP and how to efficiently deal with them when confronted with an infringement claim. This was debated by panelists who experienced both sides of the negotiation table. 

Ira Blumberg (Senior Vice President at VideoLabs – a PAE – and previously CIPE at Lenovo) explained that you basically have three options when confronted by a PAE: negotiate, litigate or ignore. Although ignoring seems to be the easy choice, Ira explained that this is usually not the best option. One of the key considerations is to analyze whether the PAE has a track record and realistic opportunity to actually litigate its patents (a risk-adjusted analysis of litigation). One of the things Ira found helpful was to ask the PAE “to help me convince my CFO to sign a deal”, which forces the PAE to feed data into your analysis. Furthermore, “the longer you are talking, the longer no one is shooting”. One of the negative assumptions that PAEs generally have to overcome is that they only assert low quality patents. Another factor to consider is that PAEs usually want smooth and continuous revenues, which can be used as a bargaining chip in your negotiations. 

In any event, digitalization and the Internet of Things (IoT) has catalyzed patent filings across the globe and along with it came the birth of companies that try to monetize patents solely with licenses. Like with SEPs, one of the key questions is whether a PAE should be entitled to an injunction. This question is all the more relevant for products that rely on thousands, if not tens of thousands of patents for their functioning. Should injunctions be available if only a minor portion of such product infringes a PAE’s patent? And should it play a role whether the suing party can simply go bankrupt with little to lose, whereas the sued company can lose an entire business? Depending on how this develops across different jurisdictions, PAEs may become more common and more influential in shaping IP. The panelists agreed that the time has come for legal systems to find a middle ground.

Copyright and Innovation

Article 17 of the Copyright Directive has created quite a fuzz in the realm of copyright. Some describe it a crucial compromise between right holders and platforms, others describe it as an incomprehensible provision that fails to close the value gap. Market fragmentation due to diverging implementations was repeatedly mentioned as an acute problem for copyright, with Article 17 being the most complex provision. Giuseppe Abbamot (Director for Media Policy at the European Commission) discussed the busy year ahead for the European Commission, with plans to issue Guidelines on Article 17. Giuseppe also mentioned a review of the Database Directive next year and the launch of the upcoming Digital Service Act. Marc Schuler (Head of IP and IT at Taylor Wessing) pitched the idea of a unitary copyright, but Giuseppe urged everyone to first focus on a proper understanding and application of the current rules.

Continuing the discussion on platforms and their role in protecting copyright, Katharina Hiersemenzel (Senior Vice President at Constantin Film) described the dilemma faced by movie producers. They are sandwiched between talent and network. On the one hand, producers need a big streaming network to distribute their content. On the other hand, movie producers must share the success achieved with a production with their talent. However, at least when working with US streaming services, there is no transparency from the streaming network to the producers and the producers do not participate in any success, making it impossible for them to share this with their talent. Katharina suggested to apply Articles 18 and 20 of the Copyright Directive also to providers vis-à-vis US streaming services in an attempt to foster innovation and creation.

We have also seen many publications over the past few years regarding copyright protection for the output generated by an AI-system. Think of the now famous Belamy painting (see here), or Project Rembrandt (see here), where an AI program was trained to subsequently create an entirely ‘new’ painting. In this regard, Dr. Julia Keim (Microsoft) highlighted the European Parliament’s recent proposal on how to regulate AI (see here), indicating that a distinction should be made between (creative) works made by humans assisted by AI – which should be able to benefit from copyright protection in the EU – and works made entirely by AI – which should not be able to benefit from copyright protection in the EU.

IP Strategy: How to Make IP a Valuable Asset

Our beloved IPKat readers probably already appreciate the value of IP. This traditionally does not apply to CEOs, who are primarily focused on "growing, selling or fixing" their company, Kevin Rivette (Managing Director of Sherpa Technology Group) suggested. However, he recognized a trend with CEOs, who are increasingly appreciating that IP can accommodate achieving these goals. According to Kevin this was catalyzed by several interesting developments such as the increased damages rewards in patent cases in the US (e.g. Apple case) with the possibility to claim enhanced damages (e.g. CISCO case), the return of “big licensing” (the Expedia patent deal), big patent transfers (Nokia, IBM, Blackberry are selling quality patents, while Chinese companies such as Xiaomi are buying patents), and the renewed interest for patents on Wall Street (e.g. funded IP litigation companies such as Beufeud). 

In order to emphasize the importance of IP with management, it is important to employ IP strategies that resonate with CEO’s goals to grow, sell or fix the company. It was suggested to put IP Key Performance Indicators (KPI’s) in place, and to periodically evaluate them with management, as well as the engineering and marketing team. By bringing these three disciplines together, potential IP protection opportunities can best be identified. Market experts can identify market demands and crucial differences with competitors that could benefit from IP protection (top-down approach). In turn, it can be discussed whether the engineers' solutions need to be protected with IP given the market circumstances (bottom-down approach). 

Several companies indicated that the maxim “quality over quantity” is key to keeping the costs of maintaining their IP portfolio under control. Selecting which aspects to protect is crucial, but has become increasingly more difficult for several reasons, Anna Wieczorek explained. The amount and complexity of the available data has increased significantly, and more and more data is Chinese (approximately 80% of the total patent applications are Chinese) with poor quality machine translations. Furthermore, many companies have been known to purposely obfuscate information in patent applications. This “information camouflage” strategy is used by companies to make it harder for other parties to extract useful information from the patent disclosure and eases the attempt to evade prior art. There is strong belief that many of the foregoing problems can be overcome by employing AI tools that can structure and evaluate the vast amount of data.

Another recurring theme throughout the conference was of course the COVID-19 pandemic. Hence, one of the sessions explored how this has affected three major brand-portfolio owners (Heineken, Philip Morris and Devialet) in their IP strategies. The pandemic has fueled an explosion of online shopping, and along with it came an increase in online infringement. The challenges described by all three companies seemed to relate to volume: more IP rights, more platforms, more infringements. In order to cope with this and keep up “infringement maintenance”, brand-owners need digital tools.

Lisa Ritchie (Senior Counsel at Philip Morris) explained that they use a third party service to monitor IP infringements, identifying tens of thousands of potential infringements each month and in some cases also taking enforcement actions on behalf of Philip Morris. Marlou van de Braak (Director Legal at Heineken) explained that they stepped away from third party infringement monitoring services and now use a digital search tool for quick knock-out searches. Their software allows them to monitor all 500 brands at nearly the same costs and is capable of linking infringements in different countries to the same provider. With the improvement of online brand protection, the panelists also expect platforms to take more responsibility. Partnerships between brand owners and online intermediaries are the future to tackling online infringement. Although China’s legal system has seen rapid developments over the past decade in terms of IP protection, the panelists agreed that it is still very difficult and inefficient to target counterfeit in China and that the focus is primarily on dissuading counterfeiters from selling online.

As to developing an IP strategy, Max Sills (Lead Open Source Compliance Attorney at Google) shared some valuable insights. According to Max, the development of any IP policy starts with the aggregation of all the decisions the company has made so far. These decisions provide the basics of your IP policy. At the same time, an IP policy is only as good as the employees who implement it in practice. Hence, Max suggests to “give employees a perspective and a belief system instead of a list of rules. No one can remember a list of rules, and few care to follow them.” IP policy has to encompass perspective and ideology. In order to be successful, you must teach the ideology, not the rules. For any company desiring to (further) develop its IP policy, it’s worth taking a look at opensource.google.com, where many exemplary documents can be found."