Handling deep fakes in real life

Imagine this – a video of someone dear to you suddenly starts circulating on WhatsApp, promoting a fraudulent investment opportunity or a video is circulated among their colleagues that contains scandalous statements they never made. The face, voice and gestures are recognizably like theirs, but the video is entirely fake.

This is no longer science fiction, and deepfake videos have now spread across the internet. Until now, the most visible victims of deepfakes have been celebrities, but deepfakes have already entered the lives of ordinary people. They are often AI-generated from images on social media or from other sources, such as photo studios.

So, what does one do in such a situation? The first step is to take the content down to limit the spread of harm. 

If the content is of a grievous nature, such as nudity, abusive content, or impersonation (which most deepfakes are), it must be immediately reported to the police and/or the national cybercrime reporting portal, and to platforms such as Facebook or X to have it taken down. These portals are legally mandated to remove such harmful content within 2 hours of receiving a report. Indian law also penalizes various offences in relation to deepfakes, such as offences against women, child abuse, identity theft, stealing a person’s credentials, etc. So, the police will investigate these matters, and on matters involving child abuse materials, they share the hash value of the content with platforms for take-down. 

After this, one must approach a court and seek a dynamic injunction to prevent the video’s circulation, typically against unidentified persons, known as John/Jane Doe orders (Ashok Kumar orders in India). If the content is prima facie harmful, courts are likely to order all internet service providers and platforms to remove it. Once such an order is obtained, it can be served on platforms, intermediaries, and, in some cases, internet service providers.

But taking down a single video is not enough. The Internet has a permanent memory, and a single judicial order rarely ends the harm, so there needs to be a cycle of repeated enforcement. The victim would need to keep a permanent watch over the internet and keep writing to various platforms once they reappear to take them down. 

Bad actors can continue to misuse identity to create more deepfake videos, and it is difficult to approach the courts every time a new video appears. In recent years, Indian courts have issued several orders restraining the hosting and dissemination of synthetic content featuring celebrities, including cases involving the misuse of their faces, voices, likenesses, and personas. 

That line of cases includes the now well-known actions brought by Amitabh Bachchan and Anil Kapoor. They evolved through cases involving public figures, where the law protects the commercial use of persona, voice, image, likeness, and public identity, often alongside performer-rights under copyright law. While one can argue that similar injunctions should extend to non-celebrities, the remedies available today remain largely performer-right-centric. 

However, in one sense, all of us are performers now. Our faces, voices, gestures, and expressions are constantly available for digital capture and can be endlessly replicated by AI. Yet Indian law still treats performers’ rights as belonging to a narrower, older world of live, recognisable performance in the arts. The ordinary citizen, whose identity can now be cloned as easily as a singer’s voice or an actor’s face, falls outside this legal imagination. That is why the deepfake problem cannot be solved solely through celebrity personality rights litigation jurisprudence.

Our copyright law still begins with the premise that the photographer owns the copyright in a photograph and therefore controls its use. That framework belongs to an earlier world in which photographs were largely taken in studios and the law could comfortably assume the subject’s consent. It is outdated in an age when a person can be photographed or video-graphed anywhere, and that image can then be fed into AI systems to generate a deepfake. AI service providers often have no way to verify who owns the copyright or the personal data rights in an image, and therefore, this construct sits uneasily with the Digital Personal Data Protection Act, which recognises a person’s right to the personal data in a digital photograph.

One possible way is to amend and clarify copyright-related rights, such as performers’ rights, to include the right of every Indian resident to claim rights over realistic imitations of their bodies and voices. This will extend copyright protections to images and likenesses, helping victims protect their digital identities online.

Denmark has already moved in this direction, proposing to amend its copyright framework to allow people to object to the sharing of realistic digital imitations of their bodies and voices, while preserving satire and parody. After all, it is not just celebrities who have the right to dignity and privacy. 

The writer is Partner – TMT at Trilegal.

Disclaimer: The views expressed are the author’s own and do not reflect the official policy or position of Financial Express.