DO DEEPFAKES COME UNDER THE AMBIT OF COPYRIGHT LAW?
Digitalization has had a great impact on the lives of human beings. It has made life much more exciting than it was in the late 1990s. Increasing scope of digitisation has also provided an opportunity for miscreants to misuse the tools afforded by the current technology. Digital disinformation is on the rise and it spreads throughout the internet in the form of morphed photos or videos. A tool that is used to create such morphed videos is ‘Deepfake’. Deepfake stems from the combination of ‘deep learning’ and ‘fakes’ and uses AI technology to merge, combine, replace and superimpose images and video clips to create fake videos or pictures that appear as authentic to the eye. The impact of deepfakes was seen during the Indian election as well as in many viral pornographic videos of celebrities. Given the high amount of unethical misuse being carried out by using the deepfake technology, this paper seeks to determine whether Copyright law can be used to combat deepfakes.
In the spirit of the above-mentioned facts, Part II of this paper would discuss what a deepfake is, Part III would answer whether deepfakes amount to Copyright infringement and Part IV would offer concluding remarks.
What is a Deepfake?
Deepfake algorithms use deep learning models such as auto-encoders and generative adversarial networks to examine facial expressions and movements of a person and synthesize facial impressions of one person to another making similar expressions or movements. Researchers have determined that deepfake content can be broadly classified into three categories:
● Face-Swap, in which a person’s face is swapped with that of someone in the original video
● Lip-sync, in which an arbitrary recording is imposed on the video of a person whose mouth movements are slightly changed
● Puppet-master, in which a target person is animated by a performer sitting in front of a camera and acting out what they want their puppet to say or do.
Deepfakes had gained publicity in 2017, when a reddit user had posted fake porn videos using existing content available on celebrities and artificial intelligence algorithms. As of September 2019, 96% of the videos regarding celebrities are non-consensual pornography. Furthermore, deepfakes have been used as a form of political campaign during the Indian elections to level allegations against the opposing candidates. The technology has been made available to the public cheaply through the ‘Fake App’ that is available on the reddit platform.
The deepfake technology is not limited to unethical misuse. The benefits provided by the technology cannot be ignored. The technology can be used to create virtual patients, which would then remove the need to share the data of real patients. It can be used to recreate famous action movie scenes, and provide live-action animation and simulative interaction. It can even transform e-commerce by allowing the consumers to have an idea as to how they would look when they wear a particular dress or product with the help of this technology. The use of deepfake technology was seen when Reuters collaborated with AI company Synthesia and created the world's first synthesized presenter based news reports. Researchers at the Samsung's lab in Moscow were able to transform Da Vinci's old masterpiece of Mona Lisa into a video, using deep learning to show her a moving figure.
Recently, major digital animation studios have used actor’s faces on another character in the film, to effectively increase the value of their story and cut-down the production costs. In Guardians of Galaxy 2 (2018), a de-aged and young version of the American star Kurt Russell was showcased. This feat could only be achieved by using the puppet master method of deepfake technology to exchange the face with the 1980s version of Kurt Russell.
Despite the positive boost provided by deepfake technology, its negative aspects cannot be ignored. A technology that does more harm than good cannot be said to be a revolutionary one. Banning them would not be the ideal solution either, so the policy framed must target the negative use of the technology by deepfakes. One of the areas where the policy-makers may get their answers is with Copyright Law.
Do Deepfakes Amount to Copyright Infringement?
A copyright confers upon its owner, the sole right over his work, which may be an invention, design, a literary or any other form of an artistic work. This implies that copyright may exist within texts, images that have been sent through the medium of the internet that would come under the ambit of ‘work’.
Under the Copyright Act, 1957, any work of visual recording or any sound recording accompanying such visual recording is the part of a cinematographic film. Generally, the author is the owner of the Copyright unless specified otherwise. Author in relation to the photograph is the photographer, and in relation to a cinematograph film or sound recording is the producer.
Section 14 states that the Copyright owner of a cinematographic film and sound recording has the exclusive right to make license of the film or to do an adaptation or a translation , including the photograph of any image forming a part of the movie. Section 51 of the Act lays down the acts that lead to copyright infringement of protected work. Thereby any person attempting to make deepfakes of cinematographic work and publish it on the internet without the consent of the owner is liable for Copyright infringement under Section 51. Where it has been established that there has been a Copyright infringement, the owner is liable to be entitled to any remedy in the form of injunction, damages, accounts and other remedies available.
Currently, deepfakes are used to fabricate videos and photographs for immoral reasons such as revenge pornography, manipulating political campaigns and to spread fake news. In light of these matters, it must be noted that even if the author passes on the rights on their work to someone else, the authors have a moral right over their work. This was discussed in the Amarnath Sehegal v. Union of India judgement, where the Delhi High Court had noted that the authors had a right to preserve, protect and nurture their creation. The author of the work shall have the right to claim the authorship of any work and recover damages in respect of any distortion, mutilation, modification to the said work, if it is used for purposes contrary to the moral intentions of the author.
In the United States, Copyright infringements have been claimed against the deepfake videos of celebrities. Copyright takedown of deepfakes made news across the globe when the fake video of Kim Kardashian was taken down due to the infringement claims raised by Conde Nast. In another claim, the rapper Jay-Z and his company Roc-nation forced Youtube to take down two deepfake videos that were using AI to impersonate his voice.
Copyright infringement may be available to the owner of the Copyright, but it still does not solve all the problems that are posed by the use of deepfakes. Copyright claims are not available to the person whose photograph is being impersonated. For example, the right to raise copyright claims may lie with the owner of the photograph and not with the person whose photograph is being taken. So, the remedy of Copyright infringement may not lie with the target and this is one issue that the policy-makers must solve. For the purposes of copyright infringement relating to deepfakes, all the parties whose interests lie in the artistic work must be allowed to raise the issue at an appropriate forum.
The rapid advancement of technology has led to the adoption of deepfake technology. Deepfake technology has its own set of harms and positives. The harms that include pornography, trolls and unethical misuse, weigh severely on the minds of those affected. People are still experimenting with the use of this AI technology and the positive uses are still being discovered . Despite the beneficial uses of the innovation in various spheres, such as news reporting, art and history and medical purposes, the negative impacts are still felt all around the globe.
It thus becomes necessary for the policy-makers to take note that deepfakes come under Copyright law but may not extend towards the target under all the circumstances. Thus, they must ensure that the Copyright enforcement options are available to all the parties who have an interest in the work and to find the right balance between law and deepfakes to ensure its proper use within the society.
Title Image: ASIS International
This article has been written by Aanish Aggarwal who is a Second year student at the West Bengal National University of Juridical Sciences, Kolkata.