​Deepfake images and videos: If Taylor Swift & Sachin Tendulkar aren't safe...

Terms like “Taylor Swift” and “Taylor Swift AI” are no longer searchable on X (formerly Twitter).

​It happened again. Last week, a flood of AI-generated images of Taylor Swift went viral on social media. While many of the explicit deepfake images have been pulled down at the time of publication, one image in particular was viewed over 47 million times before it was removed.

Terms like “Taylor Swift” and “Taylor Swift AI” are no longer searchable on X (formerly Twitter). Satya Nadella, CEO of Microsoft, which has a sizeable stake in ChatGPT developer OpenAI, has called Taylor Swift's AI images terrible and alarming, and called for action. Indeed, there have been numerous calls for stricter legislation against deepfakes, including among US politicians in what will be a momentous Election year in the America. In May 2023, Democratic congressman introduced the proposed Preventing Deepfakes of Intimate Images Act.

The Taylor Swift AI images is hardly an isolated incident. On January 15, 2023, Sachin Tendulkar took to X to disassociate himself with a video that went viral. The video features a deepfake version of Sachin promoting a gaming app (Skyward Aviator Quest) on the Apple App Store. The visuals in the fake promo video are from Sachin’s interview with Curly Tales in April 2023 and have cleverly used deepfake technology; showing 'fake' Sachin endorses the game, as one that his daughter enjoys playing.

Sachin responded to his deepfake in a statement that captures the angst of celebrities across the world who are soft targets for deepfake videos:  “These videos are fake," he said. "It is disturbing to see rampant misuse of technology. Request everyone to report videos, ads and apps like these in large numbers. Social media platforms need to be alert and responsive to complaints. Swift action from their end is crucial to stopping the spread of misinformation and deepfakes”

Sachin’s post on X prompted a swift response from the Minister of State for Information Technology – Rajeev Chandrasekhar who didn’t just thank Sachin for highlighting the deepfake menace but also reiterated the threat to the safety and trust of Indian users posed by AI. The minister’s post also mentioned that the government will be notifying tighter rules under the IT Act to ensure compliance by platforms.

This is not the first time a celebrity in India has been a deepfake victim. In November 2023, actress Rashmika Mandanna was the target of a deepfake video which showed a woman entering an elevator. The original video featured a British Indian girl – Zara Patel; her face was morphed with Rashmika’s face. In November 2023, the Centre conducted a meeting with social media platforms to ensure they regulate content and prevent circulation of deepfakes on their platforms under Indian law.  The creator of this deepfake video (a 24-year-old engineering graduate from Andhra Pradesh) has been apprehended.

The term deepfake is a portmanteau of deep learning and fake. Deepfakes involve the manipulation of facial appearance through deep generative methods and are being misused in multiple spheres, from porn to politics. Machine learning and artificial intelligence have fuelled the rise of deepfakes. Powerful AI tools now also allow semi-skilled and unskilled individuals to craft fake content with AV clips and images. Deepfake audios are also rampant. Cyfirma, a cybersecurity company revealed that there has been a 230% increase in deepfake usage by cybercriminals.

India sixth most susceptible nation

Home Security Heroes, a US-based organisation, released a research report – 2023 State of Deepfakes. According to the report, online deepfake videos are up by 550 percent since 2019 reaching a staggering 95,820. The report suggests that it takes less than 25 minutes to create a 60-second pornographic video and that deepfake pornography makes up 98 percent of all deepfake videos online. The report has identified India as the sixth most susceptible country to the threat of deepfake videos.

Rajeev Chandrasekhar’s post is a strong indication that tougher rules are coming that can protect targets of deepfakes.

As of January 30, the gaming app Skyward Aviator Quest - promoted in the Sachin deepfake - is still available for download on the Apple App Store in India. Currently regulators can invoke Section 66D of the Information Technology Act, 2000, which pertains to 'punishment for cheating by personation by using a computer resource.' Offenders may face imprisonment for up to three years and a fine ranging up to Rs 1 lakh. While this might offer some recourse, offenders can use a clear lack of legal definition to evade prosecution or action. Deepfake technology is also evolving faster than legislators can fathom. In December 2023, Sushil Kumar Modi demanded a SEBI-like regulatory body. He advocated this as a counter measure for social media platforms to curb the threat of deepfakes and synthetic videos.

In November 2023, India was among 28 countries that signed the Bletchley declaration after the AI Safety Summit. These countries formed a unified agreement on the urgent need to understand and collectively manage potential risks of AI technologies. Closer home, Indian Institute of Technology (IIT), Madras established a Centre for Responsible AI (CeRAI), an interdisciplinary research centre, to ensure ethical and responsible development of AI-based solutions in the real world. CeRAI will formulate sector-specific recommendations and guidelines for policymakers. It also aims to provide all stakeholders with the necessary toolkits for ensuring ethical and responsible management and monitoring of AI systems that are being developed and deployed.

According to Kaspersky, a firm in the cyber security space, deepfakes are still at a stage when you can spot the signs yourself. The company suggests we look for the following characteristics for a deepfake video detection:

- jerky movement

- shifts in lighting from one frame to the next

- shifts in skin tone

- strange blinking or no blinking at all

- lips poorly synched with speech

- digital artifacts in the image

But as Deepfake Technology gets better, victims are going to need better protection under Indian law and more organisations like CeRAi that can stay one step ahead of deepfake experts who misuse the power of AI.