Deepfake landscape: Experts discuss emerging threats and need for regulations

Deepfakes have emerged as a new threat to democracy, says IT Minister Ashwini Vaishnaw. The minister has initiated a plan to draft regulations to arrest the creation and virality of deepfakes. This menace has been around for a while but the technological advancement of generative artificial intelligence has given it a leg up in recent times which has come on the radar of several governments.

Vaishnaw added that all stakeholders have agreed to come up with an actionable plan in the next 20 days based on the four pillars—detection, preventing spread, strengthening reporting mechanism, grievance appellate mechanism, and awareness.

CNBC-TV18's Nisha Poddar delves into the matter with an expert panel featuring Henry Ajder, Founder of Latent Space; Jaspreet Bindra, Founder & MD of Tech Whisperer; and NS Nappinai, Supreme Court Advocate and Founder of Cyber Saathi, in the 'Big Deal' show.

Nappinai applauds the initiative but emphasises the need for an expanded approach. She suggests that watermarking, initially proposed for user identification, should also indicate that an image is a deepfake.

She said, “Well, nothing can be really foolproof when it comes to technology. But it's a good start. Now, if we just expand to a lot of the other things that he mentioned, he talked about bringing in the four pillars in terms of detection, prevention, reporting and awareness and in that, he specifically talked about the use of watermarking as a means to identify the user who uploads deep fakes so that they can introduce this concept of accountability. Watermarking should not be limited only to identifying the user, but also to the fact that it is a deep fake.”

Drawing on global legal trends, she highlights the importance of making tampering with the watermark a punishable offence.

She added, “Now, if you look at some of the trends in terms of laws in foreign countries, and specifically I would draw on New York in this, it says that one, you have to watermark it to show that it's a deep fake, and two manipulating that watermark by removing it to show it as if it's real imagery is also an offence. So something similar would be needed. We need something and we need it fast.”

Ajder, shedding light on the global scale of the problem, underscores the challenges in legislating against such harms effectively.

Ajder said, “This is a truly global problem. Research I conducted back in 2019 found that this was an issue affecting all different regions of the world, from East Asia to South America to India and indeed Europe and the US. So, this is something that's only become more and more prevalent as the tools have become commodified, they've become democratised. Whilst I think it's promising to see Indian legislators considering this approach, there are some pitfalls when it comes to legislating these specific harms and it is incredibly challenging to do so effectively, and to hold them to account.”