A recent video of actress Rashmika Mandanna went viral on social media and netizens reacted with a mixture of shock, surprise and horror when they realized that it wasn't the popular actress but a British Indian influencer named Zara Patel.
Also Read| 'Extremely scary': Rashmika Mandanna flags 'identity theft' issue as deepfake video goes viral
The actress quickly went on social media to share her despair and shock over the viral video. In her comments on X, Mandanna wrote, "I feel really hurt to share this and have to talk about the deepfake video of me being spread online."
"Something like this is honestly, extremely scary not only for me, but also for each one of us who today is vulnerable to so much harm because of how technology is being misused." the actress added.
“Deepfakes are trained with many pictures and videos to learn facial features. This program, a neural network, has two functions: creating face-swapped videos and detecting such modifications. Known as Generative Adversarial Networks, or GANs, these networks improve through a tug-of-war where one part tries to fool the other. As the system gets better, it becomes more challenging to discern if a video is genuine or altered by this intelligentsoftware.” Dr. Sunil K Shekhawat, CEO, SanchiConnect- a company that works with many startups in the deep tech sector- told LiveMint.
While Deepfakes aren't directly recognized under Indian law, the issue is indirectly addressed via Section 66 E of the IT Act which makes it illegal to capture, publish, or transmit someone's image in the media without that person's consent, CyberPeace Foundation founder, Vineet Kumar told LiveMint.
“The maximum penalty for this violation is ₹2 lakh in fines or three years in prison. The DPDP Act's applicability in 2023 means that the creation of deepfakes will directly affect an individual's right to digital privacy and will also violate the IT guidelines under the Intermediary Guidelines, as platforms will be required to exercise caution while disseminating and publishing misinformation through deepfakes. ” Kumar added.
1) Unnatural eye movement and facial expressions: “Deepfakes may exhibit eye movements that don't look natural or a lack of eye movement altogether, like an absence of blinking. Additionally, facial expressions may appear off or not match the emotion of the dialogue.” says Sunil K Shekhawat.
2) Inconsistencies in physical appearance: Shekhawat says this can include awkward facial-feature positioning, unnatural body movement or posture, and abnormal coloring or shading on the face and body.
3) Audiovisual discrepancies: “Deepfakes often have poor lip-syncing, robotic-sounding voices, or background noise that doesn't seem right. These are indicators that the audio and video components may not be authentic.” added Shekhawat.
Catch all the Business News , Breaking News Events and Latest News Updates on Live Mint. Download The Mint News App to get Daily Market Updates.