Deepfakes and deception: The looming challenge of AI in India

Deepfakes have existed for a long time now, with some of the earliest instances going back to 2017. But what once required a certain level of skill has now become a child’s play thanks to Generative Artificial Intelligence (Gen AI).

In India, Gen AI-powered deepfakes became a part of mainstream conversation when late last year a deepfake video of Rashmika Mandanna, a popular actress, went viral. Today, Mandanna is not the only celeb targeted by deepfakes, and names like Alia Bhatt, Ranveer Singh and Katrina Kaif have all had their faces pulled on fake videos.

It seems like yesterday when AI was synonymous with generating text from prompts; now, we’re already closing in on AI tools that can create hyper-realistic video and converse like real companions. Recent announcements from OpenAI and Google I/O also prove that AI will soon be able to both see and hear you using an omni-channel approach.

ALSO READ: How to prevent AI chatbots from collecting your data

It’s safe to say AI is the biggest innovation since the internet. However, just like the internet, AI too is quickly turning out to be a double-edged sword. Your kid’s homework or a cover letter now needs seconds instead of hours. But the scary side of AI is much more sinister.

Deepfakes are a quickly growing concern, especially in a country that’s already dealing with misinformation. As AI continues to get more powerful, making someone believe something fake is real, is only likely to get easier.

The need for AI regulations

“By 2027, GenAI smartphones will account for 43 per cent of overall smartphone shipments, surpassing 550 million units” Akshara Bassi, Senior Analyst at Counterpoint Research said. With AI being able to create images and videos out of just an idea, these abilities coming to smartphones also raise the likelihood of more misinformation.

While social media and instant messaging have made it easier for fake news to spread like wildfire, AI tools are making fake news more convincing.

“Deepfakes are increasingly blurring the lines between reality and fiction, with the potential to mislead public opinion, spread misinformation, and influence key events like elections. As such, regulations to address this threat are now an urgent necessity,” says Prabhu Ram, Head, Industry Intelligence Group at CyberMedia Research (CMR).

“As the ecosystem in India is still in its nascent stages, the government is striving to strike a balance between regulation and innovation,” Ram adds. This is why the need for AI regulations and policies is crucial right now. Unless we have measures to control and regulate what AI can be used for (and to what extent), the risks posed by AI may always get in the way of true innovation.

Current state of AI regulations in India, and overseas

As of today, India has no regulations over how AI is used. While this has been great for innovations in the AI space, no restrictions also mean anyone with access to tools like ChatGPT or Midjourney can use them freely, with no repercussions.

To be sure, several regulatory guidelines have been introduced over the last five years. This includes the first National AI Strategy (#AIFORALL) by Niti Ayog in 2018 and its continuation in the Principles of Responsible AI released in 2021. The government also released the DPDP (Digital Personal Data Protection) Act in 2023, which oversees the processing of digital personal data in India.

However, these guidelines have more to do with how AI uses data and how AI-powered resources contribute to various sectors. For the moment, there exist no regulations against consumer level misuse of AI, including the use of AI-generated images, voice samples or even videos.

ALSO READ: OpenAI announces a major ChatGPT privacy update

Meanwhile, usage of AI tools and services overseas is quickly being regulated. Globally, countries are taking proactive steps to combat deepfake misuse. Policies are being developed to enforce transparency, enable labelling and tracing of such content, and enforcing criminal penalties on deepfake content creators.

According to a BBC report, the creation of sexually explicit images of an adult is a criminal offense in England and Wales. While several states in the US have reportedly updated their laws to add deepfakes to existing prohibitions.

What can be done, and the way forward

In India, there are no specific regulations governing the use of AI. More updated legislation around the use of AI should be the first step India needs to work towards. Campaigns around raising the awareness of AI-generated content, and steps to fact-check images, audio and video could also help.

“While regulations are essential to provide structure and clear boundaries, especially concerning data sharing, privacy, and copyrights, the principle of ‘responsible AI’ must begin at the design level and continue through deployment,” Bassi says. “This involves transparency about training data, ensuring diversity and equity, and maintaining accountability, human oversight and continuous retraining of AI models.”

Ram adds, “A multifaceted approach, encompassing of adequate regulatory guardrails, international cooperation, and creating greater public awareness about the threats posed by deepfakes is critical.”

As citizens, we too can help minimise the spread of misinformation. Proper fact-checking when it comes to news, WhatsApp forwards or anything else that feels a little too good to be true, are great practices.

Educating those around us on what AI can do and helping them make a habit out of not blindly trusting anything that’s not verified are also great ways to help. Lastly, reporting any unverified content on social media, which could be instances of deepfakes, or other forms of misinformation is also a good way to do your part.