This article is written by Shrushti Shedge of Symbiosis International University, Symbiosis Law School, Hyderabad Campus, an intern under Legal Vidhiya
Abstract
The swift development of the Artificial Intelligence (AI) has radically changed the technological and economical frameworks, and algorithms become the core of AI-based innovation. With AI algorithms increasingly becoming more prevalent in such critical areas as healthcare, finance, governance, and law enforcement, the necessity of an effective legal protection has become quite an urgent issue. Nevertheless, the current legal systems were mainly created to safeguard human-developed works and inventions, which makes them insufficient in case of considering the special features of self-educating and autonomous AI frameworks. This essay is a critical analysis of the legal issues in AI algorithm protection in the legal environment today.
The article evaluates the suitability and shortcomings of the conventional intellectual property regimes, such as copyright, patent, and trade secret regulations, in defending AI algorithms. The study also examines how the trade secret law could be used as an alternative mechanism and its weakness with the growing pressure on algorithmic visibility.The paper also explores the nexus between AI algorithm protection and data protection and competition law, highlighting how privacy rights and antitrust issues limit the exclusive proprietary claim. Comparative analysis of the international practices shows that there was no harmonization of AI algorithms regulation in jurisdictions. The paper has ended on a note to propose a moderate and flexible system of law that incorporates intellectual property rights, regulatory controls, and policy changes to guarantee innovation as well as accountability of the developing AI ecosystem by the people.
Keywords
Artificial Intelligence, AI Algorithms, Intellectual Property Law, Patentability of Algorithms, Trade Secret Protection, Data Protection and Privacy, Competition and Antitrust Law
Introduction
Artificial Intelligence (AI) has altered modern technological and economic systems at a fast rate, and algorithms are the fundamental element of AI-based innovation. AI algorithms can allow machines to learn through data, predict and accomplish tasks that have previously been performed by human intelligence. AI algorithms are now used in such critical areas of society as healthcare diagnostics and financial forecasting, automated decision-making in governance and law enforcement. Therefore, the algorithmic protection on a legal basis has become an increasingly significant concern.
Regardless of their commercial and strategic utility, AI algorithms have serious legal ambiguity concerning ownership, protection, and enforcement. Conventional legal systems, especially those related to intellectual property (IP), did not accommodate self-learning autonomous systems, but rather human-made products and inventions. Such a misfit has created loopholes in protection and has resulted into arguments on authorship, inventorship and misuse. In addition to that, the growing dependency on the data-driven models evokes concerns associated with the privacy, transparency, and competition law.
Regulators and courts across the world are struggling to find a balance between innovation incentive and the consideration of the common good of the people like accountability and fair competition. Although there have been efforts in some jurisdictions to apply existing legal principles to AI-related works, others have suggested new, or combined, regulations. Cases of software, algorithms and data-driven technologies provide a good indication of how the law can be changed in this sphere.
The paper is a critical review of some of the legal issues surrounding the protection of AI algorithms. It examines the relevance and the weaknesses of intellectual property law, the trade secret framework, data protection laws, and competition law in reference to applicable cases laws. The paper also examines the global solutions and new legal measures to the AI algorithm dilemma.
AI Algorithms: Idea and Legal Characterisation.
The AI algorithms are designed as a coarse of commands that allow machines to manipulate data, detect patterns, and make decisions with the least amount of human supervision. In comparison to traditional software, AI algorithms, especially machine learning and deep learning, change over time as a result of being trained on massive data sets. This flexibility makes it hard to classify them legally because even the ultimate output or behaviour of the algorithm is not always predictable when it is being developed.
Legally, AI algorithms are in an unstable place. As much as they are intangible assets, they cannot lie well in the existing property categories. Algorithms have traditionally been considered mathematical mechanisms or abstract concepts, which generally do not have direct legal protection. It protects the expression of algorithms, e.g. of source code, or applications of algorithms with technical effects, instead.
In Gottschalk v. Benson, 1972, the U.S. Supreme Court ruled that a mathematical algorithm in itself could not be patented as a subject matter. Judicial interpretations with regard to software make insights on this matter. Likewise, in the case of Alice Corp. v. CLS Bank International, the Court insisted that abstract ideas that are applied using computers are not eligible unless it is an inventive idea. The approach of Indian courts has been similar as witnessed in Ferid Allani v. Union of India, where the Delhi High Court declared that computer-related inventions that have a technical contribution can be patented.
These examples suggest that the law understands algorithms not as discrete objects but mostly as functional or expressive ones. The dynamic, data-sensitive quality of AI algorithms also makes ownership and enforcement more difficult, and legal characterisation is one of the underlying issues of AI innovation protection.
AI algorithms and Intellectual Property Law.
- AI Algorithms Copyright Protection.
The traditional copyright legislation safeguards original literary works such as computer programs on one criterion that the work must be in a tangible expression and that it must be the work of human creativity. With AI algorithms, the protection of copyright typically covers the original code or the object code, and not the algorithmic logic or the ideas.
The most important issue is to meet the requirement of originality and authorship. Courts have always been of the view that copyright lasted in the expression of an idea as opposed to the idea itself. In a case of Computer Associates International Inc. v. Altai Inc., the U.S Court of Appeals has made a distinction between protectable code and unprotectable algorithms, processes, and methods of operation. On the same note, in the Eastern Book Company v. D.B. Modak, the Indian Supreme Court pointed out that the criterion of copyright protection is minimum creativity.
The problem is further complicated by the fact that AIs can write or alter code on their own. Most countries such as India have copyright laws that demand a human author. In Nova Productions Ltd v Mazooma games ltd, the Court of Appeal in the UK ruled that the computer-generated work does not have any author unless a human being is in charge of creative control. This means that the AI-generated algorithms cannot be covered by the copyright legislation, thus exposing the creators to misappropriation.
Therefore, although copyright provides partial protection of AI algorithms by way of source code, it does not sufficiently cover algorithm logic and autonomous creation which underscores a certain legal lapse.
- AI-Based Inventions and Protection of patents.
Patent law is more protective as it grants exclusive rights on inventions that meet the aspects of novelty, inventive step and industrial applicability. Nevertheless, AI algorithms experience significant challenges in patent regimes because they have exceptions on abstract concepts and mathematical procedures.
Patent offices and courts have always disallowed patents on algorithms in themselves. In the case of Alice Corp. v. CLS Bank International, the U.S. Supreme Court struck down patents covering an algorithm executed on a computer stating that only computerisation of abstract ideas is not enough. The Indian patent law also does not cover any mathematical methods and computer programmes per se in Section 3(k) of the Patents Act, 1970.
However, jurisprudence has changed to appreciate AI-related inventions that have technical uses. In Ferid Allani v. Union of India, the Delhi high court made it clear that inventions that prove to have a technical effect or contribution did not deserve to be refused a patent simply because they involve computer programs. What is still unclear, however, is the issue of inventorship especially when AI systems produce patentable outputs on their own.
The world is still willing to demand human inventors in the form of refusing to grant AI system “DABUS” inventorship in the UK, US, and EU. This strict method constrains the patent claims of AI-based innovations and highlights the insufficiency of the current patent regulations in the face of self-directed algorithms.
Protection of AI Algorithms as Trade Secret.
The trade secret law has become a viable option of securing the AI algorithms especially in cases where patenting and copyright protections are no longer an option. Algorithms find great utility in secrecy, particularly in case the competitive edge requires secrecy of model architecture, training methods, or datasets.
In trade secret regimes, secrecy is required to receive protection in case the information is commercially valuable and it is the subject of reasonable efforts to keep the information confidential. Confidential algorithms have been defended by courts in a number of cases. In Waymo LLC v. Uber Technologies Inc, the issue of autonomous vehicle algorithm theft, due to misappropriation of trade secrets successfully demonstrated that AI technologies are vulnerable to employee mobility and theft of data.
Similar recognition has been made by the Indian courts on algorithmic trade secrets as in Zee Telefilms Ltd v. Sundial Communications Pvt Ltd, the injunctions were used to secure business information that was considered confidential. Trade secret protection, however, has its own set of limitations, such as the threat of reverse engineering and non-protection in case the secrecy is compromised.
In addition, emerging regulatory requirements of transparency and explainability of AI systems are in mostly conflict with trade secret claims. This conflict is critical to the success of the use of trade secrets to ensure the long-term protection of AI algorithms.
Challenges of Data Protection and Privacy.
The algorithms used in AI are strongly reliant on large amounts of data, which can be personal or sensitive. The laws of data protection introduce responsibilities that have a direct impact on the secrecy of algorithms and the creation of the algorithm. Laws like the General Data Protection Regulation (GDPR) enforce transparency and accountability and data minimisation that could necessitate the disclosure of algorithm operations.
In Google LLC v. CNIL, the Court of Justice of the European Union emphasized balancing between the interests of data protection and business. On the same note, the case of the Dutch SyRI overturned an algorithmic surveillance system because it infringed on the principles of privacy and transparency.
The Justice K.S. Puttaswamy v. Union of India, the decision of made privacy one of the core rights that has a great impact on the regulation of AI. These rulings indicate that the privacy legislation will be able to limit the degree to which AI algorithms are kept confidential, thus making it harder to legalize them.
Antitrust and Competition Law
The dominance of the market in terms of data concentration and automated decision-making can be created or reinforced by AI algorithms. Regulators are taking a closer look at the algorithmic behavior of companies to provide anticompetitive behavior such as price fixing and collusion.
In United States v. Topkins (2015) was the court that dealt with the algorithmic price coordination as a type of cartel behaviour. The European Commission has also challenged the competition law with regards to market abuse using algorithms.
Such advancements signal challenges in the sole protection of AI algorithms with the aims of competition, especially in cases where algorithms interfere with the market fairness of competition or consumer decision-making.
International Law and Comparative Study.
The protection of AI algorithms through legal means is a matter that is understood differently in different jurisdictions. United States is quite dependent on the judicial interpretation, whereas the European Union is more inclined towards the regulatory intervention in the form of AI-specific regulations. The Artificial Intelligence Act that was proposed by the EU is based on the risk approach to accountability, but not proprietary protection.
Comparative jurisprudence demonstrates that there is no harmonization, which makes it an issue of enforcement in cross-border AI litigation. The judiciaries in different jurisdictions are coming to recognize the weaknesses of the conventional IP laws and this supports the necessity of international standards that are coordinated.
New Legal Propositions and Policy Advice
Considering the constraints of the current legal systems, researchers and policymakers have suggested sui generis protection of AI algorithms. This type of regime would be able to strike a balance between incentives towards innovation and the need to be transparent and accountable.
Judicial trends suggest that there is a change towards functional and contextual analysis as opposed to strict categorisation. Technical contribution, societal impact, and ethical concerns are increasingly taken into consideration in the AI-related disputes in courts. The complementary mechanisms that have been suggested include regulatory sandboxes, mandatory licensing and algorithm audits.
An integrated model involving the use of IP rights, regulatory controls, and contracting in dealing with the complex problems of AI algorithms seems to be the most appropriate solution.
Conclusion
AI algorithm protection has complicated legal issues that older structures do not deal with effectively. Copyright, patent and trade secret laws are somewhat inconsistent and pointless solutions and the data protection and competition laws come into place with extra restrictions.
The case law in jurisdictions demonstrates a conservative but dynamic attitude to the cases related to AI. Judicial intervention is not the solution to managing the changes in the existing laws in order to accommodate technological changes, and the judicial interpretation is important.
It requires a logical and progressive legal framework that can ensure efficient protection of AI algorithms as well as guarantee common good. The future of AI control is in adaptive, harmonised, and innovation-friendly legal frameworks that are able to respond to the swift technological shift.
References
- Alice Corp. v. CLS Bank International 2014) 573 U.S. 208 (Supreme Court of the United States).
- Computer Associates International Inc. vs. Altai Inc. (1992) 982 F.2d 693 (United States Court of Appeals, Second Circuit).
- Eastern Book Company vs D.B. Modak (2008) 1 SCC 1 (Supreme Court of India).
- Ferid Allani v. Union of India (2019) SCC OnLine Del 11867 (Delhi High Court).
- Gottschalk v. Benson (1972) 409 U.S. 63(Supreme Court of the United States).
- Google LLC v. CNIL (2019) Case C-507/17 ECLI:EU:C:2019:772 (Court of Justice of the European Union).
- Justice K.S. Puttaswamy (Retd.) v. Union of India AIR 2017 SC 4161 (Supreme Court of India).
- Nova Productions Ltd v. Mazooma Games Ltd (2007) EWCA Civ 219 (Court of Appeal, United Kingdom).
- SyRI Case (2020) ECLI:NL:RBDHA:2020:1878 (District Court of the Hague).
- United States v. Topkins (2015) No. 15-cr-00201 (United States District Court, Northern District of California).
- Waymo LLC v. Uber Technologies Inc. (2017) Case No. 3:17-cv-00939 (United States District Court, Northern District of California).
- Zee Telefilms Ltd v. Sundial Communications Pvt. Ltd. 2003(5) BOMCR404 (Bombay High Court).
- Patents Act, 1970 (India).
- General Data Protection Regulation (EU) 2016/679.
- European Union, Proposed Artificial Intelligence Act (2021).
- Abbott, R. (2020) The Reasonable Robot: Artificial Intelligence and the Law. Cambridge: Cambridge University Press.
- Bently, L., Sherman, B., Gangjee, D. and Johnson, P. (2022) Intellectual Property Law. 6th edn. Oxford: Oxford University Press.
- Lemley, M.A. and Casey, B. (2019) ‘ Remedies for Robots’, University of Chicago Law Review, 86(5) p. 1311-1396.
- Samuelson, P. (2017) ‘Allocating Ownership Rights in Computer-Generated Works’, University of Pittsburgh Law Review, 47(4), pp. 1185-1235.
- Surden, H. (2014) ‘Machine Learning and Law’, Washington Law Review, 89(1) pp. 87-115.
- World Intellectual Property Organization (WIPO) (2019) WIPO Technology Trends: Artificial Intelligence. Geneva: WIPO.
- OECD (2019) Artificial Intelligence in Society. Paris: OECD Publishing.
Disclaimer: The materials provided herein are intended solely for informational purposes. Accessing or using the site or materials does not establish an attorney-client relationship. The information presented on this site is not to be construed as legal or professional advice, and it should not be relied upon for such purposes or used as a substitute for advice from a licensed attorney in your state. Additionally, the viewpoint presented by the author is personal.
0 Comments