The rise of deepfakes: 25 percent of videos that people watch are later found to be fake

Deepfake videos are a type of synthetic media created using AI techniques, particularly deep learning algorithms. Image: Shutterstock

In the past few weeks, deepfake videos of several Indian actors have taken the internet by storm. It began with a fake video of actor Rashmika Mandanna after which videos of celebrities like Alia Bhatt were widely shared and circulated. These deepfake videos, generated by artificial intelligence (AI) tools, are deceptive, creating a challenge for viewers to distinguish between what is real and what is fake.

What are deepfake videos?

Deepfake videos are a type of synthetic media created using AI techniques, particularly deep learning algorithms. The term "deepfake" is a combination of “deep learning” and “fake”. The videos involve manipulating or superimposing existing images or videos onto source images or videos, often replacing the original content with fabricated, AI-generated material.

Listen now: New regulations to prevent deepfakes: How will they work?

The prevalence of deepfakes, especially in election-bound states, has prompted the government to take serious note of the trend, particularly within the context of the popularity of short videos. Describing deepfakes as a novel threat to democracy, Rajeev Chandrasekhar, Minister of State for Electronics and IT, has announced that the government will support citizens in filing FIRs against social media platforms for violating IT rules if they are affected by objectionable content such as deepfakes. He also indicated that new regulations will be introduced soon to address the issue. In the interim, the government has mandated that social media companies promptly and diligently identify misleading and deepfake videos. The Ministry of Electronics and Information Technology underscores the need for swift action within 36 hours of deepfake videos being posted, as specified in the Information Technology (Intermediary Guidelines) Rules of 2021.

Also read: Deepfake democracy: South Korean presidential race candidate goes virtual for votes

A survey conducted by LocalCircles, a social media platform and pollster focusing on governance, public and consumer interest issues, revealed that according to 30 percent of citizens, 25 percent of all videos they watch are later found to be fake. The survey, which includes responses from 32,000 Indian citizens from 319 districts, was conducted after the platform received a significant number of public comments and complaints related to deepfake videos in the past two years. According to the survey, 43 percent citizens watch three or more short videos on average each day over platforms such as Instagram, Facebook and YouTube, among others.

Also read: News anchors targeted by deepfake scammers on Facebook

The threat of deepfake videos is not just limited to actors, politicians or celebrities. Anyone can potentially fall victim to deepfake videos as new tools continually emerge worldwide, allowing users to generate such content. The survey says that in community discussions, individuals have voiced concerns about the inadequate responsiveness of platforms to user complaints when their social media accounts are compromised. Some have shared instances where even after filing a formal police complaint, it took seven to 10 days for deactivation, and without one, individuals are left waiting for a response from platforms even after three weeks. While the government has proposed a 36-hour timeframe for platforms to remove deepfake videos, 56 percent survey respondents feel this should be reduced to 24 hours.