Artificial Intelligence and IP: A Literature Review

With a plethora of developments and discussions on AI and AI based innovations, almost everyone has an opinion on how it should develop! While there are a number of aspects covered by “Artificial Intelligence”, ranging from definitions to scope, from life-saving to life-threatening, there has been surprisingly limited public policy discussion on the intersection of AI and IP in India. In this post, developed along the lines of a literature review cum blogpost, Yashna Walia has looked through the various government policy documents on AI to see what they have to say about IP! (For some previous posts that do discuss this interplay, check here for a post by Prof. Basheer and here for a post by Prof. Arul Scaria).

Yashna is a fifth-year law student at UILS, Panjab University, Chandigarh. Her area of interest lies in IP and corporate law.

[Long post ahead]

Image generated by DeepAI

Artificial Intelligence and IP: A Literature Review

Yashna Walia

With the launch of ChatGPT a few months back, AI made inroads into all of our daily conversations, with people even launching courses and workshops on how to use the technology, while the less-enthusiastic ones focused on the possible loss of employment it may bring along. When mentioned in the recently-held budget session, the IT and Telecom Minister Ashwini Vaishnaw acknowledged the ethical concerns and risks around AI. However, he also stated that the government was not considering bringing a law or regulating the growth of artificial intelligence in the country.

As the chatter around AI maintains its upward flight, its interplay with the intellectual property system of the country must be discussed too. As the primary policy-making body of the country, the government’s stand on it assumes significant importance. In this light, this post presents a literature review of the governmental response to AI-related IP issues and the concerns and suggestions of civil society on them.

June 2018: National Strategy for Artificial Intelligence (NITI Aayog)

Recognising that India has a significant stake in the AI revolution, the then Hon’ble Finance Minister, Shri Arun Jaitley, in his budget speech for 2018-19 directed the NITI Aayog to establish the National Program on AI. In its discussion paper of June 2018, with the hashtag #AIforAll, the NITI Aayog states that India has the potential to position itself as a leader on the global AI map. The paper details NITI Aayog’s ‘three-pronged’ approach towards achieving the same: undertaking exploratory proof-of-concept AI projects in various areas, crafting a national strategy for building a vibrant AI ecosystem in India and collaborating with various experts and stakeholders. (Page no. 5)

While the document examines 5 key sectors into account that stand to benefit the most from AI, (namely, healthcare, agriculture, education, smart cities and infrastructure, and smart mobility and transportation), it only has a few scattered mentions of IP in this context. On page 8, it mentions that India shall benefit from the AI innovation wave if it develops a robust intellectual property regime. It notes that despite the government initiatives to strengthen India’s IP regime, applying the ‘narrowly focused’ and ‘stringent’ patent laws toward AI applications remains challenging. To resolve this, it suggests establishing IP facilitation centres and providing adequate training to various authorities including the judiciary and tribunals. It is unclear what this means exactly, as on the one hand they say the “laws” are the “challenge”, and the solution is to provide training to judiciary and tribunals. Does this indicate that there is some training on how to interpret laws?

Later, on page 46, while listing the key challenges in India’s ‘march towards leadership in AI’, the report terms India’s intellectual property regime ‘unattractive’ and one of the impediments to incentivising research and adoption of AI. However, there are no studies or data about the IP system cited for this conclusion.

The report identifies algorithms and data to be key elements that ensure the operational success of AI-powered applications. On page 60, the report underlines the importance of a robust IP regime so that innovators have confidence in its application and enforcement. It states that the current IP regime poses problems in the context of generic computer programs because of the way the algorithms are designed and trained using large data sets. The report suggests that data that is fed to an AI algorithm during its training is the “key to success”, however, it is not clear how this suggestion can assist in resolving the above issue.

Consequently, on page 91, it designates ‘intellectual property’ as one of the areas wherein governmental action is required and recommends setting up a task force, comprising jointly of Ministry of Corporate Affairs and DIPP, to examine and issue appropriate modifications to the IP regulatory regime pertaining to AI. This is repeated on page 91. However, here too it is unclear what this means as only the legislature can modify the law, while the ministries / departments can at most put together committee reports or recommendations.

July 2019: Artificial Intelligence Committee Reports by Ministry of Electronics and Information Technology

Keeping in mind the possible impact of AI on the economy and society, the Government of India sought to come out with a policy framework for AI. Therefore, 4 committees were constituted by the Ministry of Electronics and Information Technology.

 1. Committee A – It proposed the establishment of a National AI Platform (NAIRP) – an open data and knowledge-cum-innovation platform. On page 4, the report stated that such a platform shall lead to, among other things, partnerships in ‘IP creation’. It states how various countries are using data and their associated IPs as a strategic asset for global dominance (refer to page 5).

2. Committee B – This report is on leveraging AI for identifying national missions in key sectors like agriculture, food, health etcetera. It does not make any mention of IP.

3. Committee C – This report is on mapping technological capabilities, key policy enablers required across sectors, skilling and reskilling, and research and development. It mentions intellectual property as a part of creating a strong R&D base (refer to page 13).

4. Committee D – This report is on cyber security, safety, legal and ethical issues. On page no. 8, it states that sharing and collaboration through open access to datasets, algorithms and new tools is beneficial for the success of cyber defenders. It however cautions that this freely available open-source software can also be used by hackers for malicious purposes who can even target entire states as well. For eg. The report at page 9 explains that AI attackers can develop ‘Adversarial AI’ which makes bonafide AI models make mistakes. Through this, they can intrude into a target system, execute attacks, affect malware and contaminate data.

The report focuses on strengthening the AI system through predictability, trust, fairness and transparency, and does not really make an elaborate mention of IP issues in this regard. There may be implications of favouring open licensing and open source developments by virtue of its emphasis on sharing expertise and experience across organisations, and the development of an ecosystem of developers, technology and end-users in the future.

February and August 2021: Responsible AI: Approach Document for India (NITI Aayog)

Coming back to NITI Aayog, after the 2018 discussion paper, the organization came up with “Responsible AI: Approach Document for India (Part 1: Principles for  Responsible AI).” The document discusses various systems and societal considerations for AI, while fleshing out the principles for responsible AI, but once again, makes only a few scattered mentions about IP. In the part discussing the system considerations for an AI, on page 24, the document notes that one of such considerations were the security risks, that come in from relying on data, the AI system’s design and the deployment environment. The document highlights that “AI systems are susceptible to attack such as manipulation of data being used to train the AI…etc.” and thus, one of the implications can be the risk of losing an IP out of a potential ‘model steal’ attack i.e. when someone steals a model by seeking outputs from the model first and then creating a replica using the model responses. In the chapter titled “Legal and Regulatory Approaches for Managing AI  Systems” (page 32) the document maps the relevant legal mechanism in place to counter these and other concerns discussed and notes that while India does not have an AI-specific law, it has the IT Act 2000, which in combination with the IT (Reasonable security  practices and procedures and sensitive personal data on information) Rules, 2011 (SPDI Rules)  provides “technology-agnostic regime for the protection of sensitive personal information.” Surprisingly, here the report does not mention any protection (or lack thereof) within the IP laws of the country.

In Part 2 of the series i.e. Responsible AI: Approach Document for India (Part 2), NITI Aayog discusses ways to efficiently adopt the principles for responsible AI, discussed in Part 1, linked above. Though the document does not discuss any specific way related to IP, it notes, on page 16, that the use of AI for consequential decision making which includes IP-related considerations for AI innovations, warrants further research.  Another interesting observation in the document was while discussing how AI can improve efficiency, it notes the Delhi High Court’s order in Tata Sky Limited v. National Internet Exchange India, where the court suggested using AI to prevent registration of identical and deceptively similar marks (page 6).

Global Partnership on AI (GPAI)

The recent joint statement (June 2023) from India and the United States mentions that both countries have committed to develop joint as well as international collaboration on AI. Additionally, a 2 million dollar grant program has been launched under the US-India Science and Technology Endowment Fund for the ‘joint development and commercialization’ of Artificial Intelligence (AI). The statement also mentions that the United States supports India’s chairmanship of the 15 member Global Partnership on AI, which India has been chairing since 2022. The GPAI Summit 2023 is scheduled to be held in December in New Delhi. GPAI experts are collaborating across four working groups and themes – Responsible AI, Data Governance, Future of Work and Innovation and Commercialisation. IP finds mention under Innovation and Commercialisation. In 2021, the GPAI experts launched GPAI IP Primer v1, which was followed by the IP Primer v2 in 2022.

Adopting the Framework: A Use Case Approach on Facial Recognition Technology (NITI Aayog – November 2022)

In its discussion paper in November 2022, called ‘Adopting the Framework: A Use Case Approach on Facial Recognition Technology’, the Responsible AI principles (released in 2021) were examined by using Facial Recognition Technology (FRT) as its first use case. On page 61, the report mentions how facial data is very economically valuable for companies developing or deploying FRT systems, and thus forms a part of their intellectual property. It is not clear what this refers to though. Economic viability is not a metric for copyrightability, and a limited metric for patentability. It could perhaps be referring to trade secrets, however, there is no trade secret specific legislation in India.  It goes on to state that just like other AI systems, FRT systems too tend to be opaque and are not easily available for independent public scrutiny. This is because of the closed nature of their datasets and code. Thus, if the data inculcated in FRT systems is biased, people who are subject to these inaccuracies face a humongous task in front of themselves to prove that they are being subject to such discrimination. In this regard, the report notes that trade secret and IP further hamper grievance redressal efforts by individuals affected by such discrimination or bias.

Other than that, while talking about accountability issues and legal liabilities, it states how FRT systems suffer from the ‘many hands problem’ (Page 19). They receive inputs at various stages of their development – be it designing the software, training the system and testing how it functions. Indian law enforcement agencies that have deployed FRT systems have refused to share details regarding the FRT system or the databases, citing protections under trade secrets and intellectual property rights; which is an issue of concern as it hampers grievance redressal. (Page 62). It should be noted that the report makes no mention of any specific kind of intellectual property rights while talking about these systems. 

Other Documents

In the report by the Ministry of Commerce and Industry’s Artificial Intelligence Task Force, on page 30, IP is mentioned under ‘enablers’ for AI entrepreneurship. The report, relying on IP trends, notes that not many patents have been filed in AI in the past few years as compared to China and the US. And therefore, emphasizes the need to have a strong IP mechanism to encourage and protect innovations in AI. It is unclear whether the report’s recommendation is focusing on increasing R&D to lead to AI innovations, or on lowering the patentability bar and increasing enforcement mechanisms. The distinction has important policy implications, so clarity would’ve been helpful on this front.

Another important document is the Generative AI Report by INDIAai, the Government of India’s National AI Portal. It recognizes that generative AI systems are trained by reading, viewing, and listening to copies of human-created works which are subject to copyright protection. The report states that there are no copyright laws right now that would provide protection to any wholly AI generated model or creation (Page no. 37). It suggests that while fair use laws do permit the use of copyrighted material without the owner’s permission, ongoing legal disputes could disrupt this status quo; which brings uncertainty to the future of AI model training.

In the above-linked report, on page no. 38, Prateek Sibal, who is a Programme Specialist of Digital Innovation and Transformation at UNESCO, wondered about those research papers or studies where Generative AI or ChatGPT are mentioned as co-authors, and what that means for original scientific work. The report thus highlights the need for there to be a legal framework that addresses the problems regarding collaborative works that involve AI as it becomes capable of generating fresh and original content.

Research and Recommendations by Industry and Civil Society

While IP may not be at the centre-stage of discussions revolving around AI right now, it has certainly caught some attention. For instance, Google, while providing its inputs on the working draft of NITI Aayog’s Responsible AI documents of 2021, mentions IP, while talking about developing a rich data ecosystem.  It underlines the importance of voluntary data-sharing frameworks to provide adequate intellectual property protections (Page 3). While talking about the limited access to India-specific datasets for developing AI tools, it acknowledged how private enterprises have made such datasets available over the years. It states that these tools and datasets have been made freely available to third parties in conformity with intellectual property laws.

Assessing the interplay between AI and IP with the help of some numbers, NASSCOM, in its paper titled ‘AI Patents – Driving Emergence of India as an Innovation Hub’, after studying the patents filed by various companies over the last 10 years (2010-2020), states that there is almost a ‘10 times growth’ in AI-related patent filing since 2012 (pages 13-17). It further notes that India ranks 8th in AI patents, and 4th in terms of AI scholarly papers. This growth was attributed to – The enterprise demand for AI capabilities towards business resiliency and augmenting human productivity. From a vertical perspective, consumer electronics & personal computing devices, and healthcare lead AI patent filings in India. Whereas, from an assignee business perspective, the technology sector leads AI patents, followed by electronics & electrical equipment industry. The paper indicates that better awareness shall lead to increased filing, including of industry specific AI patents and broad AI patents covering use cases across multiple real-life applications.

Similarly, in Responsible AI – Guidelines for Generative AI by NASSCOM, the infringement of intellectual property is mentioned as one of the potential harms associated with the research, development and use of Gen AI technologies. It obligates the developers of Gen AI solutions to strictly adhere to applicable data protection and intellectual property rules. It also obligates the users of such technology to ‘demonstrate transparency’ in claims for intellectual property (Refer to pages 6, 10 and 13).

Reform in Law

However, while the patent application numbers are on the rise, the Industry has expressed concerns over the patent prosecution and examination regime in India. Similar to many other papers, the above-linked paper (Unpacking India’s IP Ecosystem, the industry body, on pages 32 and 43, advocates for procedural changes, like fixing time limit to file a pre-grant opposition, also suggests to reconsider the requirement of Form 27 and introduce clarity and specific guidelines with respect to technology patents.

Others such concerns were flagged by the Centre for Internet & Society here (pages 10-11), which states that Section 3(k) of the Act precludes algorithms from being patented, and the Computer Related Inventions (CRI) Guidelines were controversial because they provided for the patentability of mere software without a novel hardware component. Looking at the issue in a very nuanced manner, it calls for there to be a set standard that distinguishes between AI algorithms and non-AI algorithms. It also deals with the pertinent question of whether a copy made to train AI should be considered a ‘copy’ under copyright law, that is, if such a copy would amount to an infringement. It cites the example of Google, which has developed a technique called federated learning which localises the training data to the originating mobile device rather than copying it to a centralised server. It also talks about suggestions that such copies are too trivial/de minimis to qualify as infringement.

It must be noted that the Delhi High Court in its decision in OpenTV Inc vs. The Controller of Patents and Designs and Anr. suggested to reform our patent laws to include the patentability of business method patents.  A guest post by Pragya and Lakshita questioning the reasoning / basis for this can be accessed here.


It would be apt to say that there is certainly a positive push from the side of the government towards increased innovation and entrepreneurship regarding AI, with it realising the development AI can bring across sectors; and does not shy away from a public-private partnership in this regard. The role of IP here is important, as it incentivises such efforts, however, it appears that there has not been much in-depth thought or analysis on this front. “IP” is usually used as a generic term, despite the various forms of IP (patent, copyright, trademark, trade secrets, etc) all having very different laws, impacts and implications.

However, finding mention of intellectual property in governmental documents has been quite a strenuous task. There are some major governmental documents that make no mention of IP, like the Ministry of Electronics & Information Technology’s report on Leveraging AI for Identifying National Missions in Key Sectors. Other governmental documents have sufficed with a few scattered mentions of IP at one place or the other, without sufficiently elaborating upon the alleged ‘risks’ related to IP that they seek to identify.

A recent news article mentioned that the seven working groups constituted under India’s national programme for artificial intelligence ‘INDIAai’, who have been tasked with creating a framework for data governance, setting up an India data management office and looking into the regulatory aspects of AI, can soon submit their reports to the Ministry of Electronics and Information Technology (MeitY). Similarly, the DPIIT is said to have been working on a draft new industrial policy, which has been circulated amongst various ministries for approval. It would be a delightful prospect if these reports and policies pay attention to the interface between AI and IP and put forward recommendations for suitable reform in the system, in light of the above discussion. Moreover, the Union Cabinet has recently cleared the Digital Personal Data Protection Bill, 2022, to be tabled in this session of the Parliament. It would be interesting to see its effect on AI systems and further, its interplay with the IP system of the country.

If the readers know of any other important policy documents, please do share them in the comments section below!

Tags: ,

About The Author

Leave a Comment

Scroll to Top