Dr. Pradeep Mahapatra
Deepfake technology relates to manipulation of visual and audio content of an original video to prepare a fake copy with desired alternations. Deepfake videos are created engaging machine learning and artificial intelligence tools. Such powerful softwares became available since 2018 as desktop and smartphone applications in public sphere. Launch of generative artificial intelligence in November 2022 leveraged the technology to further heights. Within five years, by the end of 2023, deepfake videos attracted attention of world leaders as a potential threat to fair election campaigning in democracies.
With advent of photography and cinematography the practice of morphing duplicates to original pictures came into practice. Introduction of analogue and digital video expanded the field. Easy availability of machine learning and artificial intelligence technologies in the twenty-first century prompted researchers engaged at higher educational institutions to experiment with creation of fake images that look natural. A user in social media ‘Reddit’ named ‘Deepfakes’ created a community during the end of 2017. Members of the online community shared deepfake videos they have created on experimental basis. It is told that such works involved celebrity faces snapped into bodies of actress in pornographic videos. As such deepfakes were created to satisfy academic interest, neither the videos were widely circulated nor questioned about.
While the first stage in experimentation of deepfake videos cornered around porno video content, the second stage involved use of audio to clone. Viewers could not instantly recognise the negative approach in presence of video and audio in scenes. But the third stage entered into use of the deepfake video technology for exploitation, intimidation and personal sabotage. It was used to spread misinformation in a make-believe format. In the fourth stage deepfake video technology expanded into election campaigning for opinion shaping that is considered a threat to democratic values.
There was widespread accusation about use of deepfake videos during the USA presidential election campaigning in 2020. In the following elections at United Kingdom, Argentina, Slovakia, South Korea etc. the menace of deepfake videos were evident. In India video clips of Bollywood star Amitav Bacchan accusing the ruling BJP chief minister and portraying contesting Indian Congress Party chief ministerial candidate in positive light was a glaring example of threat of deepfake videos in poll-bound Madhya Pradesh. Two video clips from the sets of popular television show ‘Kaun Banega Cororpati’ created political controversy during November 2023. BJP filed complaints with the Election Commission and the police.
By the same time, release of deepfake videos into the social media on Bollywood film stars Rashmika Mandanna, Katrina Kaif and Kajol created sensation in the national level. Experts cautioned that the menace of deepfake videos will speedily rise in the country. Information Technology ministry and the Prime Minister expressed grave concern on the development. The politicians in the country are skeptical on the subject as state assembly and general elections are scheduled during 2023-2024. While contemplating upon role of deepfake videos in the election campaigning, a few aspects come to the forefront:
Firstly, government calculation reveal that internet connections in India reached 85 crore by March 2023. Kantar-Google report claimed that half of the internet consumers in the country consume news from online sources. When people depend upon social media for information, naturally it will contribute significantly to opinion shaping. Use of artificial intelligence products such as deepfake videos will have a major impact on upcoming elections.
Secondly, previously preparation of deepfake videos required a huge quantity of data. But with improved technology, one to two minutes of video clip is sufficient to clone deepfake video and audio content of a particular character or a scene. Moreover software for the purpose is freely available. Anybody with practice can master the art using a simple smartphone. All such developments will contribute for production of increased number of deepfake videos in India.
Thirdly, while internet penetration in the country is in a steady pace, data rates remain one of the lowest in the international level. The environment is conducive for the people to access more video content, download and share. As a result more number of deepfake videos go viral.
Fourthly, while considering political role of deepfake videos a complete changed scenario surfaces. Previously production of deepfake videos was expensive. Only big political parties having more money could benefit from the technology. But by 2023 software in form of mobile apps are freely available and politicians in the Gram Panchyat level are capable of using the facility. Thus, use of deepfake videos may play a significant role in election campaigning during 2024.
Fifthly, it appears to be a difficult task to curb circulation of deepfake videos in the social media and encrypted messenger platforms. About 80 percent of internet consumers in the country are present in social media platforms. Social media has power to make any video viral in a short time. Though the laws and rules governing online content is robust in India, before identification of a deepfake video either by public or government to ask the social media platform to block or remove, the video in question could have already reached a substantial number of audience. Technological competence to auto-delete a depfake video seems to be a distant dream.
The history of deepfake videos by 2023 extending to past five years reveal that progress of the technology involved for production doubled every year. The number of makers of deepfake videos also grew in a competitive manner. Such a complicated situation compels governments in different countries to control the situation through regulatory mechanism. European Union and Untied States of America have taken-up preliminary steps for enactment of laws. In India, Information Technology Rules 2022 prescribes 36 hours for the social media platforms to remove an objectionable content after receipt of a complaint by the public or direction of the administration. But in the evolving online scenario 36 hours seems to be a lengthy period. Most of the objectionable content will be identified only after reaching a considerable audience.
Social media consumers play a vital role in promotion of deepfake videos. Research findings reveal that lack of online etiquette and mischievous attitude are the two main causes which lead to sharing of offensive videos. Media experts have pointed out that spread of ‘media literacy’ among the public can promote responsible online behaviour. They explain that the number of people who become aware on negative impact of deepfake videos in democracy will contribute to the number of people who keep themselves away from consumption and circulation of such harmful content.
Soon after the Madhya Pradesh incident of deepfake videos during November 2023, Prime Minister put forth his views for awareness building which technically refers to ‘media literacy’. Immediate steps in this direction is essential. Journalism Schools can play a definitive role in propagating ‘media literacy’ in their neighbourhood communities in the regional level. University Grants Commission can formulate a model with a budget for implementation of a programme which can be further extended to student volunteers of National Service Scheme. Such a campaign can enlighten a section of the society before the general election 2024.
(English translation of the original Odia newsletter by the author circulated on November 24, 2023. https://tinyletter.com/pradeepmahapatra/letters/message-328. It is an open-access content, free for translation and reproduction)
Dr. Pradeep Mahapatra is a retired faculty of Journalism, Berhampur University, Odisha.https://about.me/pradeepmahapatra
References:
Interview with Hany Farid by Ketaki Desai. Deepfakes are eroding trust. How as we even know if a real video is real ? Sunday Times of India (Bhubaneswar Edition). November 19, 2023
Mishra, Bhavini. Quick redress key weapon in battle against deepfakes. Business Standard. (Bhubaneswar Edition). November 18, 2023
