Artificial Intelligence (AI) refers to the ability of machines or computer systems to perform tasks that typically require cognitive abilities and human intelligence. Such systems are designed to replicate the logical and analytical functioning of the human brain, allowing them to consume data, take feedback, adapt to changes, and improve their performance over time without any explicit programming changes. AI systems can perform diverse tasks, including problem-solving, analytical reasoning, general learning, knowledge perception, natural language understanding, and decision-making. These characteristics give this technology immense potential but also make it possible for misuse.[1] Deepfakes is a type of synthetically generated media that involves the use of Machine Learning (ML) algorithms and models, Generative Artificial Intelligence (GAI), and digital image processing (DIP) techniques to create highly realistic but entirely fake content. Deepfakes involve using artificial neural networks – training generative networks on large amounts of data involving images, videos, and audio recordings to discern and map the characteristics of a person’s voice, facial expressions, and mannerisms. It also consists of using discriminatory networks to enhance the quality of generated content to mimic the source more closely[2].
PROBLEMS ASSOCIATED
AI is a fast-growing sector with massive potential for boosting countries’ economies. Many countries have predicted that their economy will have a large share of AI-generated work in the next few years. In India, the AI market is anticipated to grow at a rate of twenty percent in the next half-decade, which is expected to be behind China only in the matter of pace[3] and is expected to add $ 450-500 billion to our GDP in the next two years[4]. The need for innovation in this area in the present time is crucial for the growth of any country. Although these technologies are instrumental and make the work of humans easy, they also come with many difficulties. AI and related technologies present foreseeable and unforeseeable problems such as privacy violations, biases in the algorithm, and risk for manipulation of people[5]. The deepfake technology, in addition to these problems, creates fabricated content that can lead to misinformation, deception, and defamation[6]. This becomes more problematic as it can be used for fraud and harassment, directly attacking the fundamental rights of the citizens. Also, with the apex court of the country giving a green flag to privacy as a fundamental right[7], the recent incident of a deepfake video surfacing the internet involving actress Rashmika Mandanna[8], with her face being used on another person’s video, led to widespread discussion about the violations and evils that deepfake technology is capable of. What makes these deepfakes videos dangerous is the fact that they look too real, and it becomes hard for social media users to differentiate between the real and the fake. One other thing that needs to be taken into consideration is how deepfake technology makes women more vulnerable to risks, especially in a society like India where more focus is placed on the character and honor of women and deepfakes, which can proliferate like wildfire, presents a potential danger for women among other sections of the society[9]. The use of deepfake technology, especially for the purpose of pornography, which is made entirely without the consent of the women, makes this technology have a very gendered aspect[10]. Along with these, problems also arise in the arena of intellectual property as well. Ethical considerations come when various professionals and students have started working just with AI, making ideas and innovations in work sterile. Many questions remain unanswered, and ambiguity in some areas persists, such as regarding ownership of AI-generated work, who gets the responsibility if the AI-generative work infringes upon the intellectual property rights of a person, and using licensed material without permission[11].
EXISTING PROVISIONS IN INDIA
There is no specific legislation on AI in India as of now. However, there are some provisions, guidelines, and regulatory frameworks that could be helpful when considering the governance of AI in India. The provisions of the recently enacted Digital Personal Data Protection Act[12] aimed at the dual purpose of protecting the rights of the people to have the protection of their personal data and ensuring that the personal data is processed only for legally allowed purposes. This, although it does not directly address the question of AI, ensures that the data of users are not used for unlawful purposes, and this can be used to make sure that the developers using these technologies do not develop AI models using data sets that have not been consented by the users for the specific purpose and they use only legally allowed data set. Along with this, certain provisions of the IT Act[13] relating to digital deception, such as section 43 A[14] (provides for compensating individuals if there is a failure to protect their data), section 66 D[15] (which relates to cheating through the use of any computer resource by personation) and section 67[16] (punishing the publishing of obscene data online), could also be helpful. Along with these Information and Technology rules[17], which focus on the role of intermediaries and social media platforms, there are also some provisions that could be helpful in getting relief for people who are wronged by the misuse of these technologies. These, again, do not directly address the questions related to AI but could be used by people who have witnessed the downside of AI and related technologies. Also, the IT Act, when formed, was the time when the internet was in its initial stages, but as the internet has grown by leaps and bounds, the IT Act has failed to keep pace. The Indian Copyright Act[18] also has no specific provisions regarding AI, and ambiguity remains over many questions, such as authorship considerations of AI or AI violating the intellectual property rights of others. Provisions of the Indian Penal Code[19] could also be used and tweaked to address specific rights violations, such as deepfakes. Also, the government established the Ministry of Electronics and Information Technology, which is a welcoming step in the direction of developing a regulatory framework for regulating AI and related technologies. From time to time, NITI Aayog also releases guidelines and plans for regulating AI and related technologies such as deepfakes (such as the national strategy for artificial intelligence)[20]. The most recent example of swift action taken by the government would be the advisory released as a quick response to the above-mentioned Rashmika Mandanna incident, which mandates social media platforms to be diligent and take appropriate steps to ensure that there is no spread of misinformation and other harming AI-related technologies specially deepfakes[21]. The advisory also mentioned that the social media platforms are mandated to take down any data that has been reported within 36 hours, and they have to take action when cases are reported as per the IT Rules and follow the time frames specified therein. India is definitely going in the right direction, and eventually, comprehensive legislation for AI and related technologies is expected. Still, some urgent issues need to be addressed quickly, such as the highly harmful use of these technologies, as is in the case of deepfakes.
APPROACHES AROUND THE GLOBE
Due to the enormous potential that the AI sector has with its pros and cons, countries around the globe are trying to come up with ways to regulate it along with its related technologies. There mainly seem to be two broad patterns in how countries try to regulate AI and associated technologies. The focus of these regulatory mechanisms is on high-risk factors along with ethical considerations and ensuring no violation of rights[22]. Some countries have tried to come down heavy-handedly on AI and are making comprehensive and specific regulations to govern it, while some countries are giving non-binding broad guidelines and placing more faith and responsibility on the developers to pay heed to ethical considerations when developing AI or any related technology[23]. Some countries, such as China, have already passed specific provisions for AI (Algorithm recommendation management provisions)[24]. Many countries have embarked on a journey to develop AI-specific legislation; notable examples are the European Union, South Korea, and Brazil. Other countries that are more lenient with the regulations include the United States, Japan, and Singapore[25]. The approach of the US[26] could be said to be tilted more towards innovation and contains five non-binding, voluntary principles and guidelines that are considered to reduce the harmful impact of AI. These would not be accepted properly as regulations due to their nature and could be beneficial for ensuring that the AI systems when in the developing stage, bloom in full innovation and are not curtailed in the initial stage. But this could also have a downside, as leaving such an important and developing area solely on non-binding principles could be harmful to the rights of the people. On the other hand, the EU’s approach[27] is criticized by people for being too restrictive and too broad, which becomes problematic when everything is measured against the same scale. Although this approach focuses more on the rights and privacy of the users, this could hamper the innovative ideas for the development of AI[28]. As the sector is still emerging, it is too early to say which approach is better at this point in time. But one thing is for sure: AI is way too important to be left alone due to its omnipresent characteristic as it becomes a dangerous and powerful tool when it is in the wrong hands and misused.
ONE APPROACH FOR INDIA
The author of this article agrees with NITI Aayog[29] and believes that India should consistently be updating its laws to be on par with technological innovations that could be used to threaten the fundamental rights of the citizens. Continuous measures are taken in the country, such as introducing the new Digital India Act[30], recently proposed to replace the outdated IT Act. The author propounds that to ensure that innovation in the emerging sector of AI continues at a full pace and at the same time, the visible and foreseeable harms directly attacking fundamental rights are curbed and punished stringently; the government should take a middle path-taking aspects of both soft laws including regulations for developers of these technologies and hard law for punishing people misusing the same. There is a need to link the laws with reality. The government could incorporate elements of the American approach, which focuses on giving voluntary guidelines, and the European approach, which focuses on aggressive regulation. The government could issue (apart from the guidelines that NITI Aayog issues from time to time) comprehensive guidelines and develop a regulatory framework for the developers of AI, focusing especially on the need for attributing some responsibility to the creators as well as making sure that developers do take into account that development of their respective technologies and its subsequent use do not hamper with the fundamental rights of the citizens. Also, the government should ensure that these guidelines have special provisions so that the needs of the already vulnerable section of society are taken care of. At the same time, developers should also take some responsibility to develop technologies in line with existing laws and guidelines and ensure they do not use or misuse data sets that infringe upon citizens’ fundamental and intellectual property rights. Some general laws to help tackle the development of unforeseeable and specific laws to tackle the present dangers of such technologies are desirable. With this, specific laws for deepfakes and other AI-related dangers are the need of the hour. Although there are existing provisions that could help punish the wrongdoers misusing the technology but, as technology gets more sophisticated, the law needs to get in line, too. Clearly defining the technologies, such as deepfakes, and outlining the corresponding punishment is required. Along with this, to ensure development and innovation in AI, instead of putting restrictions on the developers in the initial stages, the government should make stringent rules for testing the AI technologies before their release, especially when the technology is in a vital sector and involves using sensitive user information. There is also a need to ensure informed consent for the users of the technologies along with social media, and the data shared by people on the web, even when shared in a public forum, is not used without their permission and not misused. Also, in line with other countries, India should have clear and specific provisions to ensure that AI systems developed do not have some inherent biases and see elaborately that the already vulnerable section of society is not targeted or discriminated against using these technologies. Strict and clearly defined laws are also required in areas where AI is used to violate citizens’ fundamental rights and privacy.
CONCLUSION
As the AI sector is still in the developing stage and has great capability to boost the economy of any country and improve the quality of life of all, innovation in this field is essential. However, every coin has two sides, and so does this technology, which could pose a danger to the life and rights of the people of this world due to its misuse, as happens in the case of people making deepfakes, injuring the reputation of people, or spreading misinformation. AI is far too important to be left alone, and the need for regulation is real. As countries try to tackle AI using different approaches, it would be too early to know which approach is best. India needs to find its own approach to AI, and the author has propounded one such approach that the country could follow. Some general guidelines with stringent testing before launch and specific legislation for specific harms, such as deepfakes, are the need of the hour. Also, India needs to ensure that innovation happens in the sector and that we are on par with global standards. The ethical considerations of AI need to be addressed to ensure that the technology with vast potential is used only for the good of humanity and not as a tool to destroy humanity, especially the already vulnerable section of the society.
[1] Artificial Intelligence (AI) | Definition, Examples, Types, Applications, Companies, & Facts | Britannica. (22 Nov. 2023). https://www.britannica.com/technology/artificial-intelligence.
[2] Ian Sample, ‘What Are Deepfakes – and How Can You Spot Them?’ The Guardian, (13 Jan 2020). https://www.theguardian.com/technology/2020/jan/13/what-are-deepfakes-and-how-can-you-spot-them.
[3] Radha Basu, ‘The Rising Potential of AI/ML in India’s Growth Story,’ The Times of India. (Accessed 23 Nov 2023). https://timesofindia.indiatimes.com/blogs/voices/the-rising-potential-of-ai-ml-in-indias-growth-story/.
[4] Ministry of Electronics and Information Technology, India AI 2023, First edition by expert groups (October 2023).
[5] What Are the Risks of Artificial Intelligence (AI)? Tableau, (Accessed 23 Nov 2023). https://www.tableau.com/data-insights/ai/risks.
[6] ‘How Are Deepfakes Dangerous?’ University of Nevada, Reno, (Accessed 23 Nov). 2023https://www.unr.edu/nevada-today/news/2023/atp-deepfakes.
[7] Justice K.S. Puttaswamy and Anr. vs. Union of India (UOI) and Ors, (2017) 10 SCC 1 (India).
[8] ‘Rashmika Mandanna, Deepfakes, and Us: I Find Myself Constantly Questioning the Safety of Uploading Personal Information and Disclosing My Travel Location.’ The Indian Express, (18 Nov 2023), https://indianexpress.com/article/opinion/columns/rashmika-mandanna-deepfake-social-media-9032088/.
[9] Anupriya Dhonchak, ‘Rashmika Mandanna Deepfake: Are We Protecting Vulnerable Users Online?’ The Indian Express, (9 Nov 2023), https://indianexpress.com/article/opinion/columns/rashmika-mandanna-deepfake-are-we-protecting-vulnerable-users-online-9020607/.
[10] Id.
[11] Gil Appel, et al. ‘Generative AI Has an Intellectual Property Problem,’ Harvard Business Review, (7 Apr 2023). hbr.org, https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem.
[12] The Digital Personal Data Protection Act, 2023, No. 22, Acts of Parliament, 2023 (India).
[13] The Information Technology Act, 2000, No. 21, Acts of Parliament, 2000 (India).
[14] Section 43 A, Id.
[15] Section 66 D, supra note 14.
[16] Section 67, supra note 14.
[17] The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 (India).
[18] The Copyright Act, 1957, No. 14, Acts of Parliament, 1957 (India).
[19] The Indian Penal Code, No. 45, Acts of Parliament, 1860 (India).
[20] NITI Aayog, National Strategy for Artificial Intelligence #AIFORALL (published in June 2018)
[21] ‘Govt Advisory Asks Social Media Firms, Intermediaries to Axe Deepfake Content,’ The Economic Times, (8 Nov 2023), https://economictimes.indiatimes.com/tech/technology/govt-sends-advisory-to-social-media-companies-asks-them-to-take-action-on-deep-fake-content/articleshow/105037649.cms.
[22] Mikhail Klimentov, ‘From China to Brazil, Here’s How AI Is Regulated around the World’, Washington Post, (5 Sept 2023). http://www.washingtonpost.com, https://www.washingtonpost.com/world/2023/09/03/ai-regulation-law-china-israel-eu/.
[23] Id.
[24] ‘How Countries around the World Are Trying to Regulate AI’, Quartz, (11 Oct 2023), https://qz.com/how-countries-regulate-artificial-intelligence-1850916279.
[25] Id.
[26] O’Shaughnessy, Hadrien Pouget, Matt. ‘Reconciling the U.S. Approach to AI’, Carnegie Endowment for International Peace, (Accessed 23 Nov 2023) https://carnegieendowment.org/2023/05/03/reconciling-u.s.-approach-to-ai-pub-89674.
[27] Id.
[28] Meredith Broadbent, What’s Ahead for a Cooperative Regulatory Agenda on Artificial Intelligence? Csis, (2021).
[29] Supra note 21.
[30] Ministry of Electronics and Information Technology, Proposed Digital India Act, 2023 (published on 9th March, 2023).
Author: Karnica Bhatia
Juris Centre 