Deepfakes worry Indians: 1 in 2 surveyed say they watched multiple fake videos; 7 in 10 want them criminalised

According to the survey, one in two Indians said they had watched multiple fake videos in the last 12 months, while seven in 10 believe creating deepfake videos should be a criminal offence.48% Indians say over 10% of videos they watched were fake.Rapid advances in artificial intelligence (AI) have ushered in a new era of technological disruption, with deepfakes emerging as an increasingly serious challenge. A LocalCircles survey of more than 35,000 people found that one in two respondents said that more than 10% of the videos they watched on social media, OTT platforms, or WhatsApp in the past year were later found to be fake.The survey also found strong support for stricter punishment. Seven in ten respondents said creating deepfake videos should be made a criminal offence.Survey comes as Centre plans separate law on deepfakesThe survey comes as the Centre is considering a Bill to make digital arrest scams and AI-generated deepfakes distinct criminal offences. The move follows the Supreme Court's observation on Tuesday that existing laws do not specifically deal with such offences.Solicitor General Tushar Mehta informed the court that the proposed legislation would classify AI-generated deepfakes as a separate crime with specific legal consequences.The proposed law follows the IT Rules that came into force on February 20, 2026. These rules require social media platforms and other intermediaries to clearly label AI-generated content and remove illegal synthetic media within a fixed time'A new threat for society'The Central government has vehemently flagged the growing menace of deepfakes as a 'new threat for the society'. AI-generated videos, images and cloned voices are increasingly being used to spread misinformation, damage reputations, commit financial fraud and manipulate public opinion.Globally, the number of deepfake files is estimated to have increased from around 500,000 in 2023 to nearly 8 million in 2025.The survey quoting a Thales report stated that 65% of organisations had experienced deepfake-related attacks, while 55% reported losses due to AI-generated misinformation. A McAfee study estimated that nearly 90% of Indians had encountered fake AI-generated celebrity endorsements, with victims losing an average of ₹34,500.Supreme Court calls for dedicated lawDuring suo motu proceedings on digital arrest scams, a bench comprising Chief Justice Surya Kant and Justices Joymalya Bagchi and V. Mohana noted that while the court has taken a strict stance against digital arrest frauds, including denying bail in such cases, there is no specific penal law addressing these offences.Addressing Attorney General R. Venkataramani, the bench said, “The government may need to formally define digital arrest in penal laws.”The judges further observed that digital arrest scams involve elements of extortion, robbery, and organised crime, and asked whether the government should create a standalone offence with stringent penalties, including provisions to freeze the accused's assets.Justice Bagchi said framing and enforcing new criminal laws fall within the government's domain."We now have deepfakes, which can be used to mislead people or for cheating and impersonation. Using our exclusive powers under Article 142 of the Constitution, we can neither define a crime nor create a new offence," he said.About the AuthorKanishka SinghariaKanishka Singharia is a Senior Content Producer at Mint with a passion for news, trends, and the stories shaping the digital world. She specialises in spotting viral narratives by constantly tracking social media platforms and turning them into engaging, reader-friendly stories. Her work ranges from fast-paced breaking updates to sensitive human-interest features, blending speed with clarity. With over four years of experience in news and trend reporting, Kanishka has worked with leading organisations such as Hindustan Times and Times Now. She moves seamlessly between profiling business leaders and t