The etymology behind the term ‘deepfake’ is a combination of ‘deep learning’ and ‘fake’. A deepfake is an artificial intelligence-based synthetic media that replaces individuals in existing images and videos with another person’s likeness. Generally, deepfakes take the form of videos and pictures that are highly realistic but may also be in the form of audio and text.[i] Deepfakes are generated using deep learning and generative adversarial networks whereby a process of generation and discrimination is repeated until the new synthetic data that is created most closely resembles the original data on which it was based.[ii] The generation of deepfakes in this manner does not require any particular skill with modern-day technology and tools that are available publicly and at low or no cost.
Deepfakes are not in themselves harmful. It is the manner of their utilisation that gives rise to concerns. For instance, in February 2024, an employee of a multinational firm was tricked into paying fraudsters a whopping USD 25,000,000 (US Dollars Twenty-Five Million) through the use of deepfake technology impersonating the voice of the company’s chief financial officer.[iii] Deepfakes are generally targeted at prominent celebrities, people in business and politicians as it increases the probability of belief and trust being reposed by the common person in the media that they consume. Some crimes that may be committed using deepfakes are identity theft, forgery, defamation, hate speech, election manipulation, cheating, cyber terrorism, violation of privacy, obscenity and pornography, and spreading misinformation and hatred.
The rising use of deepfakes became prominent in India during the 2024 national elections. Deepfakes were used both to spread misinformation and hate as well as to communicate effectively with the diverse demographic in India. On one side, fake videos of famous Bollywood stars and politicians were being circulated in support of or against a particular political party, and on the other side, the Prime Minister of India was making use of artificial intelligence to translate his campaigns into regional languages in real time.[iv] The problem surrounding the regulation of deepfakes comes from this. Deepfakes have immense potential to be used to make false claims, cause reputational damage and aid crime, and deepfakes are invasive of privacy and undermine people’s trust in others including regulatory and governmental authorities. On the other hand, deepfakes also improve accessibility, communication and creative expression, and are beneficial in education, film entertainment and forensics. This is why the effective regulation of deepfakes becomes critical, as it will reduce the misuse of emerging technologies while ensuring that the growth of such technologies is not impeded or inhibited.
Currently, there is no specific law in India to address the concerns surrounding deepfakes. Without such particular legislation and policies, the most effective way to deal with deepfakes is to exercise additional vigilance while consuming digital media. Some standard methods to detect deepfakes include (a) observing mismatches of the audio and visual content indicating manipulation; (b) observing visual artefacts such as missing details and mismatched or unnatural movements arising from limitations in deep learning; (c) implementing deep learning algorithms trained to distinguish between real and fake media; and (d) analysing patterns and features and undertaking forensic based detection.[v]
Any misuse of deepfakes generally falls within the ambit of the Information Technology Act, 2000 (“IT Act”), the Information Technology (Reasonable Security Practices and Procedures and Sensitive Personal Data or Information) Rules, 2011 (“SPDI Rules”), Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 (“Intermediary Rules”) or the Bharatiya Nyaya Sanhita, 2023 (“BNS”). Section 66C to Section 67B of the IT Act provides punishments for identity theft, cheating by personation, violation of privacy, cyber terrorism, transmitting obscene or sexually explicit material, etc. The SPDI Rules provide for protections about sensitive personal data such as passwords, financial information, health and medical information and records, biometric information, etc. , the Intermediary Rules provide for the duties and obligations of intermediaries with regard to due diligence and grievance redressal in the digital realm. The BNS is the primary substantive criminal law in India that legislates for how crimes such as fabrication of evidence, forgery, cheating, defamation, provocation of breach of peace, etc., are dealt with. Additionally, the Representation of the People Act 1951 caters to elections- specific situations such as corrupt practices including undue influence and promoting enmity or hatred between different classes of people. The Copyright Act of 1957 provides for remedies for copyright infringement. Further, the Ministry of Electronics and Information Technology has issued statements and advisories reiterating the responsibilities of intermediaries in the face of the threat of deepfakes.
However, the current framework described above is not consolidated, and the laws only provide for general criminal sanctions that are not specific to artificial intelligence. It was hoped that India’s latest law in the realm of data protection, the Digital Personal Data Protection Act, 2023 (“DPDP Act”), would address this matter as it was an effort by the Indian government to meet international data protection standards. However, while the DPDP Act obliges data fiduciaries to undertake limited, purpose-driven, consent-based processing, it does not protect publicly available data. This means that the rights and protection available to publicly available digital media against manipulation through artificial intelligence are limited.
Furthermore, the current laws are not proactive in preventing the misuse of deepfakes and are more focused on addressing issues after they have occurred. They do not require adequate disclosures from media generators and intermediaries, leaving users unaware that they are viewing artificially generated content. This lack of transparency is a significant issue for user and consumer protection, as it can lead to potential harm. Additionally, the current avenues for grievance redressal are time and resource consuming , and place the burden of proof on the victim to establish the occurrence of such a crime.
For these reasons, India must consider having specific legislation that addresses the use of artificial intelligence holistically. It has been reported that the Government of India is planning to introduce a bill titled the ‘Digital India Bill’ that will delve into the use of artificial intelligence and digital content generated through its usage.[vi] It is to be evaluated whether this bill will adequately address critical issues about deepfakes. Separately, India must also allocate sufficient resources to incentivise the development of detection algorithms and countermeasures. It must sufficiently educate the masses and spread awareness amongst the public about the issues surrounding deepfakes. These measures will help India keep pace with the technological advances that are developing at an unprecedented pace.
Authors: Vishnu Nair, Chinmayi Venkatesh
Publication Date: September 27th, 2024
[i] M. M. El-Gayar, et al., A Novel Approach For Detecting Deep Fake Videos Using Graph Neural Network, 11 Journal of Big Data (2024), https://journalofbigdata.springeropen.com/articles/10.1186/s40537-024-00884-y.
[ii]Samuel H. Silva, et al., Deepfake Forensics Analysis: An Explainable Hierarchical Ensemble of Weakly Supervised Models, 4 Forensic Science International: Synergy (2022), https://www.sciencedirect.com/science/article/pii/S2589871X2200002X?via%3Dihub.
[iii]Heather Chen and Kathleen Magramo, Finance Worker Pays Out $25 Million After Video Call With Deepfake ‘Chief Financial Officer’, CNN World (February 4, 2024), https://edition.cnn.com/2024/02/04/asia/deepfake-cfo-scam-hong-kong-intl-hnk/index.html.
[iv]Meryl Sebastian, AI and Deepfakes Blur Reality in Indian Elections, BBC News (May 16, 2024), https://www.bbc.com/news/world-asia-india-68918330.
[v]Samuel H. Silva, et al., supra 2.
[vi] Aishwarya Paliwal, Government to bring Digital India Bill to check deepfake content: Sources, India Today (June 15, 2024), https://www.indiatoday.in/india/story/digital-india-bill-narendra-modi-governmnet-deep-fake-videos-rajeev-chandrasekhar-2553674-2024-06-15.
