A video of Rashmika Mandanna (apparently), wearing a low neck, strappy black bodysuit, entering a lift went viral.
However, users soon pointed out that it was a deepfake video, where the actor’s face was morphed into a video of a different woman. Actor Amitabh Bachchan expressed concern, and said it was a strong case for legal action.
The original video is supposedly of Zara Patel, a British-Indian influencer who had posted this video on Instagram.
Afterwards, a supposed video of actor Katrina Kaif started doing the rounds. Towards the end of November, a deepfake video of Alia Bhatt, which showed her face morphed onto a girl wearing a blue co-ord set making obscene gestures went viral. Users also commented how such videos of actor Kiara Advani can also be found. Actor Kajol has also been a victim.
Deepfake is the use of advanced forms of machine learning, to manipulate and morph video content for it to resemble other people. It leads to manufactured videos of people saying or doing things that they haven’t done in reality.
This latest use of AI and technology to create videos of people raised serious questions about privacy, safety and fake information.
HerZindagi spoke to experts to decode this new technology and what individuals can do to safeguard themselves online.
How Deepfakes Are Created
To understand the nuances of deepfakes, and the technology used in them, HerZindagi spoke to Amit Jaju, a global cybersecurity expert.
Amit explained, “Deepfakes are created using sophisticated artificial intelligence techniques such as deep learning and generative adversarial networks (GANs).”
The problem of manipulated images where faces are morphed onto bodies of other people has existed for a long time now. But, deepfakes use more in-depth technology, which makes them more complicated than those photographs.
Why Deepfakes Are More Sinister
While ordinary filters and apps can help change faces of people or swap them, deepfakes are way more sinister.
Comedian Tanmay Bhat used such an app to make caricature videos of singer Lata mangeshkar and cricketer Sachin Tendulkar and landed in controversy. However, the videos were very obviously him, and nobody mistook them to be the people he tried to impersonate.
What makes deepfakes stand out is how accurate they are. It takes a lot of extra attention to discern whether the video is real or not.
“They differ from photoshopped images in that deepfakes can create entirely new content that is often indistinguishable from real footage, whereas photoshopped images typically involve editing existing photographs,” said Amit.
A content creator Chris Ume rose to fame using his striking similarity with Actor Tom Cruise, which he used to create deepfakes. These videos are similar enough for any ordinary person to mistake them to be of the actual actor.
The purpose such technology is used for also makes deepfakes scarier, with the possibility of it being used to create graphic or pronographic content.
How to Spot a Deepfake Video
The first step of reporting any deepfake video is being able to spot a deepfake video, given that they’re eerily ‘real’.
Amit explains that spotting a deepfake can be challenging, but there are tell-tale signs. “Unnatural blinking, mismatched lip-syncing, inconsistent lighting, and poor-quality edges are some signs to look out for,” he said.
In the widely circulated Rashmika Mandanna video, if one looks closely, there is a point where she’s just about to enter the elevator where one can spot the face changing.
“Advanced methods include using software to analyze the video for inconsistencies or even AI-based detection tools,” added Amit.
How to Protect Oneself From Becoming Victims of Deepfake Videos
With so much of our data easily and openly available on the internet, the biggest concern for individuals is how to remain safe.
Sunny Nehra, founder of SecureYourHacks, a cyber security company posted on X (formerly Twitter) about one should be careful with how much people share online.
“The more the photos of yours available on the internet, the more realistic your deepfake would be. This is why the deepfakes of celebrities are so realistic. If you are very concerned about your deepfakes getting created, avoid or limit sharing photos, videos in public domain,” he wrote.
Amit, the cybersecurity expert, added that managing the privacy settings available on social media applications is crucial to keeping oneself safe.
“To protect oneself from becoming a victim of deepfake videos, individuals should manage their digital footprint carefully, using strong privacy settings on social media, and being cautious about sharing images and videos online. Awareness and educating oneself about the existence and nature of deepfakes is also key,” he said.
What Should You Do If You Find Your Deepfake Online?
The first step one must take if they find a video of theirs online, is document it and report it to the hosting platform, explained Amit.
“An individual may need to contact cyber crime authorities,” said Amit. “In India, one can report such incidents to the cyber cell department of their city's police force or the Indian Computer Emergency Response Team (CERT-In).”
One may also need to file a police complaint, explained Mudit Kaushik, Partner at Verum Legal, specialising in IPR law.
“Police complaint is needed if there is a privacy breach or defamation,” he said. “Platforms are obliged to adhere to IT Rules 2021, which includes taking down such content expeditiously.”
As directed by a recent advisory, platforms are required to expeditiously remove deepfakes within 36 hours as mandated by the IT Rules 2021, with non-compliance potentially leading to legal action under the IPC.
In the Rashmika Mandanna case, Union Minister Rajeev Chandrasekhar tweeted that the platforms can be taken to court if they don’t comply with the IT rules.
What Laws Do Deepfakes Fall Under?
Deepfakes essentially fall under several different legal arenas.
Mudit, the IPR lawyer, explained, “In India, deepfakes are subject to several legal areas including privacy, intellectual property, information technology, and criminal law. The IT Act, 2000 tackles identity theft (Section 66C), impersonation (Section 66D), privacy invasion (Section 66E), and obscene content (Sections 67, 67A).”
Furthermore, the main crux of deepfakes hinge on defamation, given the nature of such videos.
“Defamation issues are managed under Section 500 of the Indian Penal Code (IPC) for reputational harm,” he said. “Forgeries through deepfakes could fall under Section 468 of the IPC.”
There is also copyright infringement involved the creation and circulation of deepfakes.
“When deepfakes incorporate someone’s image, performance, or creative work without permission, it may constitute a breach of copyright under Section 14 of the Copyright Act,” said Mudit.
What More Can Be Done To Safeguard People Against Deepfakes?
Digital literacy is key, agreed the experts. That is the main factor that helps individuals keep themselves safe, know their rights and also distinguish between fake and authentic content.
“Investing in AI and digital forensics research is vital for developing more sophisticated detection and prevention tools, keeping society a step ahead of deepfake threats,” said Mudit, the lawyer.
Amit, the cybersecurity expert, highlighted the need for legislation to catch up with the ever-evolving deepfake technology.
“India needs comprehensive laws addressing digital impersonation and privacy,” he said. “Education and awareness at all levels of society can also serve as a first line of defense, equipping people to navigate and respond to the challenges posed by deepfakes.”