les Nouvelles September 2020 Article of the Month:
Patent Law Based Concepts For Promoting Creation And Sharing Of Innovations In The Age Of Artificial Intelligence And Internet Of Everything
Boehmert & Boehmert
Lashmikumaran & Sridharan
1. The Role of the Patent System in the Age of AI/IoE
When the Venetians created the first patent system in the 14th Century, the role thereof, as it is widely believed by scholars, lawyers and users of today's patent systems, has been to encourage inventors not to keep their inventions secret anymore, rather to disclose them through a governmental channel to the public. That would, according to the intention of the creators of the first patent system and still believed to be applicable today, give "followers," i.e., "improvers" of certain technologies that otherwise would have to be held secret to stand, as Isaac Newton has expressed it, on the shoulders of giants etc., thereby improving the speed of technological developments.
Meanwhile, in the high-tech-area of which Artificial Intelligence (AI) and the Internet of Everything (IoE) are characteristic features, practically everything is disclosed rather rapidly by creators of technical innovations, i.e., "inventors," be it by internet, be it intentionally, be it unintentionally. The need for a system encouraging inventors to disclose, instead of keeping certain developments secret, has therefore dramatically diminished.
Accordingly, nowadays modern scholars believe that the main role of the patent system is no longer to encourage inventors to disclose, rather to solicit and steer, respectively, capital investment into Research and Development (R&D), which choice investors will make only if they can protect the results of innovative activities in a reliable manner against unauthorized use, so that a marginally higher profit can be expected by investors into R&D compared with e.g., buying rental apartments, doing other investments, etc. As a consequence, the main purpose of the patent system, is to enable innovators to create a fence around their innovations preventing unauthorized use of their inventions and therefore enjoying a competitive advantage, leading to higher prices, profits, etc., obtainable by their products, compared with competitors.
Another function of the patent system, however, deeply embedded into it at least since the coming into force of TRIPs, is derived from Article 7 thereof, namely, to enhance transfer of technology and therefore the possibility of "non-inventors" to enjoy the benefit of innovations created by investors into R&D and, thereby, becoming e.g., patent owners.
In other words, a strong patent system, amongst other IP systems, is not only meant to have the effect of steering investment into the creation of innovations, but there should also be a "door" in the aforementioned fence, namely in order to enable third parties to enjoy the fruits of innovative activities, once created. The aforementioned "door" in the "fence," being an integral part of the patent system as defined by TRIPs, being the base law, so-to-say, of all patent systems in the modern high-tech-world, needs to be opened under fair, reasonable and non-discriminatory provisions to third parties that did not participate in the creation of innovations, but wish to enter the protected "garden," defined as the area covered by certain innovations, against a fair contribution to the development cost, sharing their profit with the originators.
The respective "entrance fee," with regard to the aforementioned door, by the way, has not necessarily to be paid in cash (only?), rather e.g., by cross-licenses, duly valued, as part or even complete component of the entrance fee.
The modern patent systems, like in Germany or India, make sure that not only a reliable fence, consisting of patents, can be built around innovations, thereby fulfilling the investment-steering function as described above, but contain also provisions for enabling sharing of such innovations with others, particularly "newcomers" to certain fields of technology, as will be described below.
2. Specific Challenges for the Patent System in the Age of AI /IoE
The function of the patent system as described above in Section 1, does not raise specific problems
in areas where usually a single patent, or few patents only, cover a whole product or a limited number of products. This is the world of e.g., pharmaceuticals. In that regard, it only needs to be made sure by suitable legal provisions—such as compulsory licensing,—if parties in the field, i.e. potential licensors and potential licensees, cannot "independently" agree on reasonable arms-length licensing conditions, that still a sharing of the respective technologies can take place in such cases.
In this article/paper, the aforementioned single-patent world, however, will not be dealt with in any additional detail. Rather reference is made insofar e.g., to an article "Price Differentiation and the Conundrum of Exhaustion Principles," written by one of the authors of this paper and published in "Differential Pricing of Pharmaceuticals inside Europe," ZERP, Volume 58, ISBN 978-3-8329-4280-9.
In the field of AI and IoT, however, specific problems arise insofar as usually many patents, i.e., whole patent portfolios, cover certain technologies. As a specific example for this, though the problem is not restricted to that field, is telecommunication. Beyond that field, however, also other software-related inventions and their protection by patents causes specific problems, which, however, can be solved by patent-system-based provisions, as it will be shown below.
In other words, under the principles as developed by the Boards of Appeal (BoA) of the EPO, adapted also by German Courts later on, like particularly the Xth Civil Senate of the German Federal Court of Justice (GFCJ), there is no specific problem, concerning eligibility involved in protecting inventions directed to AI. This even applies to inventions made by AI. AI-related inventions are under the present Guidelines of the European Patent Office (EPO) nothing else than Computer Implemented Inventions (CII), i.e. they are patentable insofar as they are related to the technical solution of a technical problem. Similar technical requirements exist under Indian patent law in accordance with the presently applicable Guidelines for Computer Related Inventions (CRIs), which have a lot of similarities with the Guidelines of the EPO as well as the case law of the GFCJ in Germany.
At the end of the day, all such inventions—i.e., CIIs, including AI-related inventions, and CRIs—are eligible for patenting under the European Patent Convention (EPC), and under Indian Law, as long as they are directed, as already stated, to a technical solution of a technical problem. Specific questions, like who should be the inventor of inventions made by AI, including questions of remuneration for employees' inventions, need to be solved, but such questions do not cause problems with regard to incentives for creating innovations in the respective field and sharing the results of such innovations with others.
Rather, two different problems are created and exist in AI/IoT related inventions, namely as follows:
2.1.1. Scope of Protection of AI Related/Created Inventions and Accessibility Thereto
If AI-created inventions are generously patented, the problem exists that further developers of applications of the respective patented AI as well as improvers thereof will have to respect the possibly rather broad, even "prophetic" patents granted on such inventions. This is a problem not unknown to users of the patent systems in other fields, like in the case of so-called reach-through claims covering research tools in pharma discovery and development.
The authors of this paper, however, are of the opinion that the aforementioned consideration should not restrict the rather generous granting of patents on AI-related inventions, provided that access to the respective technologies, for the purpose of further developments, is given to newcomers without undue problems.
One should keep in mind, however, that the covering of "AI-produced" inventions will increase the number of patents in AI-related fields dramatically, including Standard Essential Patents (SEPs).
Another problem with regard to AI-related inventions, particularly "generously" granted ones, besides how newcomers and improvers can get access to patented technologies, is created by so-called Standard Essential Patents (SEPs), which, if really essential for using the respective technology, must be used by other players in the field, not belonging to the originators of the respective Standard, particularly newcomers.
In a nutshell: How can modern patent systems make sure that AI/IoE-related inventions, in an age where huge portfolios covering certain technologies, like telecommunication, will be developed, get access for "using" such technologies for developing further improvements on the basis thereof? Additionally, how can it be guaranteed that SEPs can be used by competitors, particularly newcomers to the respective technology, under fair, reasonable and non-discriminatory, so-called FRAND, conditions?
Fortunately, the patent systems already contain provisions which solve the aforementioned problems, as will be shown in the following.
3. Solutions proposed by Patent Systems in order to create Freedom to Improve and Freedom to Use AI Inventions, including SEPs, under FRAND Conditions—Germany and India as an Examples
3.1. The De-Blocking Provision of Article 24 (2) of German Patent Act (GPA)
The German Patent Act (GPA) contains in its Article 24 (2) an interesting de-blocking provision, namely as follows: Whoever makes an "important" improvement of a patented invention and patents it, is, in case that the owner of the "superior" patent subject matter protected by which is improved by the "newcomer," does not give a voluntary license to the "newcomer," entitled to get a "compulsory" license under the "superior" patent. That specific "compulsory" license comes at a price, however: the license under the "superior" patent is granted to the newcomer in exchange for a cross-license under the newcomer's patent, so that the combination of the license under the "superior" patent and the "dependent" patent leads to the effect that both patent owners are entitled to use the respective inventions. It is not necessary, in this case, to check public interest, as for a "normal" compulsory license in accordance with Article 24 (1) of German Patent Act, rather it is assumed by the legislator that there is always a public interest to make important innovations available to the public, which can be only done by a "de-blocking" provision of the aforementioned or a similar kind.
An elegant way, in other words, to make AI inventions that are patented accessible to newcomers, like developers of certain applications of the respective AI, will be that the newcomer develops and patents the improvement and thereafter requests from the original, superior patentee, who may be in the possession of a too-broad, prophetic patent, a compulsory license against a cross-license. This provision could also be used, by the way, in cases where research-tools in pharma would be patented in such a manner that reach-through claims would be allowed, since in that case a too-broad protection could be avoided by a cross-license of the aforementioned kind.
With certainty, provisions of the aforementioned kind, which do exist in some, but not all European countries, but as realized already by Article 24 (2) GPA for Germany, would open a reliable possibility to the public to make sure that innovations created in fields which are heavily patented, even by some kind of reach-through claims, could be further improved and then be used, under reasonable, shared conditions, by the respective "players" in a mutually satisfying manner.
3.2. The De-Blocking Provisions According to Indian Law and Practice
Under Indian Patents Act, 1970, there exist provisions that create a balance between the exclusive right granted to a patentee to prevent others who do not have the consent of the patentee for making, using, offering for sale, selling, or importing the patented invention, while making certain patented inventions accessible by way of compulsory licencing as well as by "de-blocking" an otherwise "blocking" patent. The basis for the latter is founded in the well-recognized principle of India Patent law that the patented inventions have to be worked in India on a commercial scale and to the fullest extent that is reasonably practicable without any undue delay.1 As has been discussed previously, de-blocking finds application, for example, in case of a newcomer's patent on an improved version of an AI invention over an existing AI patent, where the existing AI patent takes on the role of a blocking patent. By way of de-blocking, the technology in the existing AI patent is made accessible to newcomers, like developers of certain applications of the respective AI, so that the newcomer patentee is not hindered, due to infringement on the existing AI patent, from exercising his positive right to commercialize and practice the improved AI invention patented in India.
Section 91 of the Patents Act, 1970 expressly enshrines the de-blocking provision under the Title "Licensing of related patents." It is pertinent to note that Section 91 is contained in Chapter XVI of the Patents Act, 1970 which is titled (and pertains to) "Working of Patents, Compulsory Licences and Revocation." It is interesting to note that the deblocking of AI patents for example, can come in the form of the deblocking provision under Section 91 or even in the form of a Compulsory Licence under Section 84. However, Section 91, specifically, empowers a person who is unable to work his patented invention because of a blocking patent, to apply to the Controller of Patents at the Indian Patent Office for grant of a licence for the blocking patent, essentially to deblock said patent, on the ground that he is prevented or hindered without such licence from working his invention efficiently or to the best advantage possible.2
Paralleling the practice of reciprocation and cross-licencing as followed in jurisdictions such as the EP jurisdiction, an applicant, such as a newcomer possessing an improved AI patent, who seeks such a licence for a blocking patent is required to be able and willing to grant a cross-licence in respect of his invention on reasonable terms to the holder of the blocking patent.3 In the context of India, however, an additional criterion exists, which requires the applicant of such a license to establish to the satisfaction of the Controller4 that the improvement patent and invention has made a substantial contribution to the establishment or development of commercial or industrial activities in the territory of India. Under the Indian law and practice, when the aforesaid requirements are met, a deblocking licence for the blocking patent may be granted and a similar order for the improvement patent if so requested by the proprietor of the blocking patent5 may be granted. The provision also ensures that such licences granted are non-assignable.
In contrast, the alternative to the deblocking provision discussed above is the grant of a compulsory licence under Section 84. A compulsory license is an authorization to any party seeking to work a patented invention in India, on the grounds that (1) the reasonable requirements of the public with respect to the patented invention have not been satisfied; (2) the patented invention is not available to the public at a reasonably affordable price; or (3) the patented invention is not worked in India. Therefore, while the fundamental basis of grant of the compulsory licence under Section 84 is public interest and welfare; for granting the deblocking licence under Section 91, it is the promotion of trade and industry in India—a purely commercial objective.
Thus, Indian law and practice allows for deblocking of the blocking patent, when (1) it hinders the working of the improved invention efficiently or to the best advantage possible; and (2) the grant of a license to the blocking patent significantly contributes to the establishment or development of commercial or industrial activities in India. In closing, the de-blocking provision serves as an effective way of fostering the working of related inventions and encourage a "patent pool" that allows the patentees to jointly exploit the patented technologies to their respective benefits. India has, however, not seen any case as yet and lacks jurisprudence in this regard.
3.3 The De-Blocking of Improvement Inventions in Other Countries
In addition to Germany and India, as indicated above, also under the laws of other countries there do exist de-blocking provisions similar to Article 24 (2) of German Law.
This applies not only to Russia, but also, e.g., to Taiwan, where in Article 87 of the Patent Act it is stipulated that compulsory licenses where a later invention or utility model patent can't be granted, e.g., "where a later invention or utility model patent cannot be exploited without infringing upon a prior invention or utility model patent, and where the later invention or utility model patent involves an important technical advancement of considerable economic significance in relation to the prior invention or utility model patent."
Accordingly, de-blocking provisions might be the key, enabling the commercial use of improvements developed on "broadly" patented e.g., AI inventions, if combined, by the way, with a generous experimental use clause, to enable the development of respective improvements.
3.4. Access for Newcomers to Standard Essential Patents (SEPs)
The German Patent Act (GPA), however, does not "only" contain, in its Article 24 (2), a provision which gives the possibility to de-block situations in which dependent inventions otherwise could not be used by anybody in a reasonable manner (which is of particular importance in the case of rather broadly patented AI inventions, but not restricted thereto), but also contains a provision that would give the famous and widely discussed FRAND declarations requested from owners of SEPs a more reliable fundament.
In this context, an approach should be considered which already since 2011 and 2012 has been discussed in Germany, particularly with and within the Ministry for Economy and Energy ("BMWi"), but also with the Ministry of Justice and Consumer Protection ("BMJV"), namely as follows:
The patent laws of many European countries, amongst them Germany (namely in the form of Art 23 (4) of GPA), provide for the determination of reasonable royalties—in case the parties that have registered a License Of Right (LOR) for a patent are not able to agree on reasonable license terms—by the respective patent office, like GPTO. Therefore, if FRAND-obliged owners of SEPs would be compelled to register LORs for their SEPs before they could ask for injunctional relief in patent litigation procedures, the burden of determining FRAND royalties would be shifted from the ordinary courts into patent offices, like GPTO.
In Germany, for that purpose, mechanisms as developed by the Board of Arbitration for Employees Inventions at GPTO could be used. The Board has a close connection to industry because of the possibility of parties in employee-intervention disputes to enlarge the Board of Arbitration by an industry representative. In order to enable the aforementioned enlargement, a pool of licensing experts of industry is designated by the GPTO in cooperation with industry in advance. In specific cases, where one of the parties of an arbitration procedure concerning remuneration for employees' inventions, particularly on the basis of license analogy, requests an enlargement, the industry expert is taken from the pool. But also in other cases, where no enlargement is sought for by one or both of the parties of an arbitration procedure at the Board, the Boards usually seems to obtain advice, concerning e.g., market-accepted royalty rates in various industries, from the pool of experts. In other words, the Board of Arbitration for employees' inventions at GPTO and its decision practice, respectively, are connected to market/commercial reality by the expertise obtainable from members of the pool of industry representatives.
One could enhance the LOR system by increasing its attraction for SEP owners by granting them easier burden of proof obligations in case of patent enforcement, namely by shifting the burden of proof that a certain SEP is not used by e.g., a newcomer to a standard to the defendant.
It is remarkable that the "Final Report" of the EU Commission on "Patents and Standards," (Ref. Ares2014)917720—25/03/2014), does mention LOR "solutions" briefly, without, however, mentioning the aforementioned possibility to have FRAND royalties determined by patent offices. This is even more remarkable because also a forthcoming Unitary Patent (UP) system would provide for an LOR mechanism similar to the German one, with appeal possibilities against royalty proposals of the EPO to the Unified Patent Court (UPC). Regulation (EU) No. 1257/2012, where in Article 8 LOR is mentioned to be filed at EPO, furthermore to be found in the Agreement relating to UPC, Article 32, under (1) (h), according to which the authority of UPC for licenses according to Article 8 of Regulation (EU) No. 1257/2012 is mentioned.
The above-mentioned mechanism of shifting the burden of determining FRAND royalties to agencies like GPTO and EPO, respectively, or to semi-private arbitration organizations would, from the viewpoint of the authors, enhance the access to new technologies, particularly for small and medium-sized entities (SMEs) that do not have large patent portfolios available for cross-licensing arrangements.
4. Patent Pools a Solution for SEPs?
The problem of royalty-stacking in case of multiple SEPs necessary for accessing a certain standard could be, according to the author's opinion, largely avoided if from presently used procedures like at ETSI, a mechanism similar to AMPEC-LA would be generally imposed onto owners of SEPs, like in telecommunication, but not restricted thereto. This would mean that the pool members would bring their SEPs into a pool under the "control" of a central administrator who has the right to grant "package licenses" under all necessary SEPs to interested parties, like newcomers.
That "package license" would necessarily have to be FRAND compliant as a whole. In other words, the overall license should not be so high that it factually would prevent any product under that "package license" to practically become un-marketable because of a too-high overall royalty load. The aforementioned viewpoint of considering the maximum royalty load of products being protected by multiple patents is explicitly considered e.g., in the Guidelines for Calculating Remuneration for Employees' inventions in Germany based on license analogy, all this being subject to Article 19 of the Guidelines. In the European and particularly German IP community discussions are on their way as to whether models similar to collection societies as used in collective copyright systems might be useful also for the distribution of the overall FRAND license fees. This would make it necessary, of course, to determine the mutual value of value/contribution of the individual SEPs and SEP packages coming from different SEP owners. With regard to the aforementioned model of "collective societies," reference is made to an article on this problem written by Christian Czychowski and Martin Schaefer and published in GRUR 2018, 582.
With regard to the role of the central administrator of a patent pool, it should also be envisaged that SEP owners would be obliged to give a license, with a right to sub-license, to this central administrator. That license, in accordance with Article 15 of GPA, would stick to the respective patent, so that even after sale of that patent to e.g., a Patent Asserting Entity (PAE) the respective patent would remain available for newcomers wishing to enter the standard by obtaining a respective license from the Administrator. This would avoid problems existing e.g., under German law with regard to "simple" FRAND declarations, which are not considered as license agreements and therefore do not guarantee to the public the same protection as Article 15 of GPA, according to which the licenses under patents covering Germany, once granted, would "walk" with the patent to any new owner, obliging that owner to stick to the original license offer to the central administrator and therefore giving, to the public, a much more reliable "guarantee" to access the respective technology than "simple" FRAND declarations can do.
5. Essentiality Check and Arbitration
It should be duly considered whether owners and right-holders relating to SEPs, but also owners of "broad" AI-related patents covering "prophetic inventions" should not be requested to let self-declared SEPs first undergo an essentiality check at a "suitable" authority, like Japan has now provided for SEPs at the Japan Patent Office (JPO), before such patents could be used in litigation, specifically with regard to injunctional requests. One could make such an essentiality check. The condition for the right to request injunctional relief from defendants, which request otherwise would be possibly considered by the courts as being an abuse of an alleged dominant position even if it would in reality not exist in a strict sense.
Similarly, one could think of making arbitration compulsory before any court action could take place, specifically with injunctional requests (of specific importance in countries with compulsory bifurcation, like Germany …), again with the consequence that only after such arbitration a respective court action, particularly with injunctional request, would be considered as lawful.
In this context, one might consider the ongoing efforts and already available mechanisms as achieved by the WIPO Arbitration and Mediation system, which is dealing with questions of SEPs in specific detail.
The WIPO mechanisms have, insofar, according to the author's experience, the advantage of containing certain, rather reasonable caps for fees, not only of the arbitrators, but also of the parties/lawyers involved on the side of the parties, and therefore could make the burden for newcomers to SEP protected technologies easier to bear than under other arbitration systems.
It is recommended that the already existing conflict solution mechanisms of the patent system should be more thoroughly used to solve specific problems of patent use and commercialization in the age of AI and IoE. The multiplication of patent portfolios, compared with the present status, will otherwise lead to tremendous problems in the world of AI and IoE, where the number of patents, particularly SEPs, but also of rather broad patent protection, will otherwise possibly impede further technological developments. De-blocking provisions and cross-licensing regulations as embedded in patent systems of certain countries, like in Germany, together with well-known arbitration mechanisms, might be a way into a more peaceful patent world in which innovators can have a smoother life than in a world of increasing global patent disputes. ■
1. Section 83 (a)
2. Section 91 (1)
3. Section 91 (2)(i)
4. Section 91 (2) (ii)285
5. Section 91(3) 286
Available at Social Science Research Network (SSRN): https://ssrn.com/abstract=3470208