RTBS Hot Topic: Deepfakes

Are deepfakes all bad, or can they be used for good?

From fans and members of the White House expressing outrage at the explicit deepfake images of Taylor Swift on X, formerly known as Twitter, to deepfake videos of PM Lee Hsien Loong and DPM Lawrence Wong used for scams closer to home, seeing is no longer believing. While deepfake technology is hardly new, the increasing misuse and disruption by bad actors raises concerns among lawmakers and the public about the real-world threat to society.

CNA. (2023, December 29). PM Lee urges vigilance against deepfake videos of him promoting investment scams. Retrieved 2024, February 26.

What are deepfakes?

Deepfakes are digitally altered content, typically videos or audio recordings, created by using artificial intelligence (AI) to realistically modify the appearance, speech, or behavior of individuals, often portraying them as saying or doing things they never actually did. The term combines deep from AI deep-learning technology, and fake, which refers to the content that is not real.

CNA. (2023, Dec 1). Deepfake scams on the rise in Singapore. Retrieved 2024, January 26.

The misuse and abuse of this technology is further intensified by the surge of generative AI and user-friendly apps that have made it more accessible to anyone. The fears of cybersecurity, scams and fraud are valid and deepfake technology has made it all the more difficult to fight misinformation and disinformation.

So, is there any good use of deepfake technology?

Photo by Google DeepMind from Pexels

Deepfakes serve as a powerful illustration of how generative AI has the potential for good across different fields such as e-commerce, healthcare, art, education and entertainment. In the world of advertising, it can help break down linguistic barriers as in the case David Beckham’s “Malaria No More” campaign where the famous footballer “delivered” the messages in 9 different languages. Hollywood is already using “de-aging technology” to bring back performers who have passed on or achieve realistic depictions of their younger selves. There is also the promise of making education more engaging, interactive and accessible through deepfake in the classrooms. Similar AI-generated video technology used in deepfakes for hoaxes and bogus content has also been used as a form of entertainment particularly on social media (see “deeptomcruise”).

According to a Bloomberg article, the potential of deepfake technology has attracted serious investments from venture capitalists (VC) with USD$187.7 million invested in 2022, up from USD$1 million in 2017. This trend is expected to continue as more use cases emerge. New York-based startup, Runway (its technology was behind the Oscar-winning film Everything Everywhere All at Once) secured USD$141 million funding in 2023 not just from traditional VCs but from tech giants Google, Nvidia and Salesforce. Additionally, deepfake media detection startups have also seen investments in 2023: Reality Defender raised USD$15 million while Deepmedia signed a USD$25 million contract with the U.S. Department of Defense.

Tackling the problems of deepfake is not easy. In addition to developing more detection tools, lawmakers across the world are proposing bills and exploring new regulations, and social media companies are updating their policies. Others have looked to public education and equipping people with critical thinking skills to safeguard the digital world.

CNA. 2024, January 10. How to report deepfake scam ads on social media. Retrieved 2024, February 26.

So, do you think that deepfake technology can ever make a positive impact?

Yes, deepfake technology is being used for good.	No, deepfake technology is far too riddled with problems.
Deepfakes can be used for educational purposes. Deepfake technology can be used for educational purposes. Museums have brought “art and history to life” as in the case of Dali Lives exhibition and Samsung’s AI research lab’s Mona Lisa. There are positive use cases of deepfakes in the classroom such as engaging with historical figures.	Deepfakes can mislead and misinform, undermining public trust. There is a growing number of deepfakes of political figures such as US President Biden, Slovakia’s election candidates discussing how to rig votes and Ukraine President Zelenskiy’s call to lay down arms. These are but the tip of the iceberg of the waves of disinformation arising from deepfake technology. Experts and lawmakers caution that deepfakes could harm reputation and disrupt elections globally.
Deepfake could potentially improve healthcare and accessibility. Generative adversarial networks (GAN), the algorithms behind deepfakes, could assist in the accuracy of cancer and dental diagnosis and therapy. Patients who suffer from voice loss could possibly preserve or reclaim their voices with deepfake audio. There is also the possibility of removing language barriers for entertainment, communication and education purposes.	The use of deepfakes is worsening online harms against women. While edited images or video are not a new concept, AI technology like deepfake has made the process easier and more accessible. Non-consensual creation of graphic images and videos are putting girls and women at increased risk of online harms, as in the cyberbullying case of young girls in Spain and harassment of celebrities like Taylor Swift. Some believe too that with more women becoming victims of deepfake, “we face a future where women’s voices are traded and their voices muted”.
Entertainment industry stand to gain from the use of deepfake technology. There are countless applications of deepfake technology in the entertainment industry – from recreating scenes with unavailable actors or depicting them younger as in the case of the ‘de-aged’ Mark Hamill in The Mandalorian and Harrison Ford in Indiana Jones and the Dial of Destiny to exploring new creative expressions. Production companies could save money and time by leveraging on deepfake technology to edit or alter videos without requiring a reshoot. Celebrities or influencers could also create personalised messages or increase their reach and presence globally as seen in footballer Lionel Messi’s partnership with PepsiCo, which enabled fans to generate customised video messages using his deepfake avatar in 8 different languages.	Deepfake scams are the latest addition to generative AI scams. Scammers can use deepfake technology to create voice recordings to trick victims into transferring money, or produce promotional videos that realistically feature celebrities or politicians endorsing products and services that they never did. In fact, in January 2024, it was reported that YouTube has deleted more than 1,000 deepfake advertisements featuring renowned personalities. Experts are concerned that as such technologies go mainstream without strong guardrails, we will see an increase in the volume and customisation of such cybercrimes.

Recommended Resources

Videos

MSNBC. (2024, Jan 26). Is 2024 going to be the ‘deepfake’ election? Retrieved 2024, January 26.

Channel 4 News. (2023, July 8). How do we prevent AI from creating deepfakes? Retrieved 2024, January 26.

TED. (2023, May 19). The Incredible Creativity of Deepfakes – and the Worrying Future of AI. Retrieved 2024, January 26.

Websites

Axelrod, J. (2024, Jan 18). Think deepfakes are bad? ‘Cheapfakes’ are far more dangerous. The Daily Beast. Retrieved 2024, January 25.

Leong, D. (2024, Jan 17). Commentary: PM Lee’s deepfake video and the risk when seeing is no longer believing. CNA. Retrieved 2024, January 25.

Katira, Kirti. (2023, Dec 27). Explained. What are deepfakes and how to spot them? Wionews. Retrieved 2024, January 26.

Patterson, D. (2023, October 5). Deepfakes for good? How synthetic media is transforming business. Tech Informed. Retrieved 2024, January 24.

Podcasts

The New Yorker. (2023, November 15). We’ve been wrong to worry about deepfakes (so far). Retrieved 2024, January 29.

Northwestern University’s Roberta Buffett Institute for Global Affairs. (2023, May 21). Breaking Boundaries: The Future of Deepfakes with V.S. Subrahmanian, PhD. Retrieved 2024, January 29.

NPR. (2023, April 27). AI-generated deepfakes are moving fast. Policymakers can’t keep up. Retrieved 2024, January 29.

BBC. (2022, May 17). The future will be synthesized: Deepfakes for disinformation. Retrieved 2024, January 29.

NLB eBooks

	Deepfakes: The Coming Infocalypse Scnick, N. (2020). Deepfakes: The Coming Infocalypse. New York: Grand Central Publishing. Retrieved from OverDrive. (myLibrary username is required to access the eBook).
	Deepfake. Abbit, Linda. The conscious caregiver: A mindful approach to caring for your loved one without losing yourself. New York: Adams Media, 2017. Retrieved from OverDrive. (myLibrary username is required to access the eBook).
	The Coming Wave: Technology, Power and the Twenty-First Century’s Greatest Dilemma. Suleyman, M. and Bhaskar, M. (2023). The Coming Wave: Technology, Power and the Twenty-First Century’s Greatest Dilemma. New York: Crown. Retrieved from OverDrive. (myLibrary username is required to access the eBook).
	Exploring Deepfakes: Deploy powerful AI techniques for face replacement and more with this comprehensive guide. Lyon, B. and Tora, M. (2023). Exploring Deepfakes: Deploy powerful AI techniques for face replacement and more with this comprehensive guide. Birmingham: Packt Publishing. Retrieved from OverDrive. (myLibrary username is required to access the eBook).