Sample A 1

Seeing Isn't Always Believing: Unraveling the Perils of Deepfakes in Politics

In a world where reality can be twisted at a click of a button, the rise of deepfakes has

ushered in a new era of digital manipulation. Deepfake is a product of artificial intelligence (AI)

that superimposes content from one video to another, consequently allowing users to create a

fake video of a person saying or doing almost anything (Ray). From the playful antics of

Snapchat's face swap filters to the jaw-dropping spectacle of Rodrigo Duterte's authoritative

voice effortlessly emanating from the lips of Hollywood heartthrob Brad Pitt, and even the most

unlikely of collaborations coming to life with Kanye West belting out a Taylor Swift ballad on

TikTok, the possibilities are both captivating and unsettling. As deepfake technology continues to

advance, it poses significant challenges to the integrity of information, individual reputations,

and democratic processes. With its ability to generate visually indistinguishable media from

genuine sources, deepfake technology has the potential to be exploited as a tool for interfering

with electoral processes, casting a shadow of doubt over the very foundations of our democratic

systems.

The political realm witnessed the initial attention surrounding deepfake technology in

2018 when a comedian impersonating Barack Obama delivered a PSA video highlighting the

deceptive nature of deepfake technology (Christopher). According to Deeptrace chief executive

Georgio Patrini, now that political leaders are becoming targets of deepfakes, this technology is

currently making a notable impact on the political landscape (Agence France-Presse).

Furthermore, Patrini points out the emergence of new cases involving the use of synthetic voice
 Sample A 2
audio and images depicting fictitious individuals, which are employed to carry out social

engineering tactics against businesses and governments (Agence France-Presse). These

developments raise concerns about the potential exploitation of deepfakes to manipulate voters,

particularly as elections approach in various countries, including the United States.

Deepfakes hold the potential to manipulate political content, paving the way for the

dissemination of misinformation and disinformation that can taint the essence of democracy

itself (Klein). In the United States, deepfake videos of President Joe Biden's speeches have

become popular, featuring the president discussing topics ranging from hip hop to drugs and

video games. However, some individuals are also utilizing deepfake technology to spread

disinformation, as exemplified by a deepfake video depicting Biden harshly criticizing

transgender women (Klein). The video, shared on Instagram on February 25 of this year, featured

a fabricated Biden stating, "You will never be a real woman," followed by a torrent of criticism

targeting transgender women, including a mention of suicide. The voice in the video was

remarkably similar to Biden's, creating an illusion of authenticity. A day after the release of this

deepfake video, PolitiFact, a fact-checking nonprofit operated by the Poynter Institute, debunked

the said video, revealing that no evidence could be found of Biden making such remarks in

White House transcripts. Instead, evidence was found of the president's efforts to support

transgender individuals, contradicting the fabricated words in the deepfake video (Cercone and

Czopek).

Deepfake videos that aim to manipulate political content and propagate misinformation

and disinformation have also emerged in the Philippines. During last year's national election
 Sample A 3
campaign, a deepfake video surfaced, featuring the late dictator Ferdinand Marcos Sr., allegedly

questioning the moral values of individuals who would elect a tax evader as president (Vera

Files). Vera Files, a fact-checking news organization in the country, debunked the deepfake video

by exposing that Marcos Sr.'s mouth had been edited and "replaced with someone else's and

made to move to deliver the remark” (Samson). The manipulated video animated an Associated

Press file photo of Marcos Sr. taken in 1986, creating the illusion that he was uttering the words:

“What is unity if your people have no moral values? Halimbawa, ‘yung kapitbahay mo ay tax

evader, tumakbong presidente, binoto mo. Anong klaseng tao ka?” One netizen commented,

“THE STRONGMAN HAS RISEN FROM HIS GRAVE!!! LISTEN!!! … (am having

goosebumps…he had premonition?)” (Vera Files).

Beyond mere fabrication, deepfakes can be strategically engineered to undermine the

stance and credibility of politicians or political parties, eroding public trust and sowing seeds of

doubt (Ray). In 2019, a deepfake video featuring US House Speaker Nancy Pelosi, in which she

appears impaired, circulated widely on social media, amassing over 2.5 million views on

Facebook alone (CBS News). Another popular Facebook post, with 91,000 shares, carried a

caption exclaiming, "This is unbelievable, she is blowed out of her mind, I bet this gets taken

down!" However, in 2020, the Reuters Fact Check team revealed that the video had been

significantly slowed down to create the false impression of Pelosi appearing drunk and

incoherent (Reuters Staff). Although the "Drunk" Nancy Pelosi video was later exposed as a

fake, its intent was clearly to cast doubt on Pelosi's abilities and undermine her credibility as a

political leader.
 Sample A 4
The danger extends even further, as deepfakes can be meticulously tailored to target

specific voter groups, deftly swaying their opinions and shaping their voting decisions (Klepper

and Swenson). On February 7, just a day before the Legislative Assembly elections in Delhi, two

deepfake videos featuring Manoj Tiwari, the President of the Bharatiya Janata Party (BJP), were

widely circulated on WhatsApp. These videos, which appeared to show Tiwari criticizing the

incumbent Delhi government led by Arvind Kejriwal, gained significant attention. One video

featured Tiwari speaking in English, while the other showcased him speaking in the Hindi dialect

of Haryanvi. In both versions, Tiwari urged viewers to support the BJP and vote for a change in

the form of a Modi-led government (Christopher). This deepfake video featuring Tiwari had a

clear objective: to undermine the credibility of the incumbent government and sway voters

towards supporting the BJP. By fabricating these videos and presenting them as genuine,

deepfakes can effectively exploit the vulnerabilities and biases of specific voter groups,

amplifying their impact and potentially swaying election outcomes.

With each new instance of a political deepfake, the trust in politicians and politics at large

further dwindles, leaving a lingering sense of skepticism and suspicion (Ray). However, despite

the widely-recognized negative impacts of deepfakes in the electoral process, there are

viewpoints that seek to undermine their significance. Some argue that deepfakes have not

emerged prominently in politics, suggesting that their influence may be overhyped or

exaggerated (Mak and Temple-Raston). Additionally, proponents of technological solutions

emphasize that deepfakes can be effectively regulated and mitigated through advancements in

technology alone, placing faith in advancements that can combat their harmful effects. On the

other hand, there are those who point out that the public has already become increasingly
 Sample A 5
skeptical and cautious regarding the authenticity of online content. A ConsumerLab report

reveals a growing trend of skepticism, with less than one in five individuals trusting the

information they encounter on social media platforms (Shyrokykh). This implies that there is

already a level of wariness and critical thinking among the public, potentially acting as a natural

defense against the influence of deepfakes. These opposing viewpoints, however, do not

overshadow the significance of recognizing the tangible risks posed by political deepfakes.

While some argue that deepfakes have not prominently emerged in politics, the

demonstrated potential of deepfakes to deceive and manipulate is undeniable. This was

exemplified by the 2018 video featuring Barack Obama speaking words he never uttered, serving

as a stark reminder of their ability to distort reality and sow confusion (Mak and Temple-Raston).

Furthermore, deepfakes played a role in the 2020 US Presidential election, where they were used

to target President Joe Biden and spread false narratives (Ray). Amidst these instances, experts

and researchers have warned about the looming threat of deepfakes in upcoming elections. Data

scientist Dominic Ligot expressed concerns about the potential impact of deepfakes on elections

in an interview with #FactsFirst, stating that the Philippines is just one election away from facing

the challenges posed by these manipulated videos (Esguerra). This sentiment reflects the

growing awareness of the potential harm that deepfakes can inflict on political systems and

societies at large.

Furthermore, despite claims that technological solutions alone are adequate in effectively

regulating and mitigating the risks linked to political deepfakes, the intricate and rapidly

evolving nature of deepfake technology renders relying solely on technological advancements an

forensics expert at the University of California, describes the rapid progression of deepfakes,

noting how quickly they have evolved in a matter of months (Galston). This rapid advancement

creates a significant disparity between the generation of deepfakes and the detection of

deepfakes. Even if reliable methods or technologies for detecting deepfakes exist, their

effectiveness will be considerably slower compared to the creation of these deceptive videos

(Galston). Consequently, false representations can dominate the media landscape for prolonged

periods, ranging from days to weeks. David Doermann, the director of the Artificial Intelligence

Institute at the University of Buffalo, emphasizes the speed at which misinformation can spread,

using the analogy that "a lie can go halfway around the world before the truth can get its shoes

on" (Galston). Additionally, Farid highlights the resource imbalance, with the number of

individuals working on video synthesis far outnumbering those developing detection techniques

by a ratio of 100 to 1. This underscores the lack of attention, funding, and institutional support

dedicated to identifying fake media compared to the creation of deepfakes, as reported by The

Washington Post (Galston).

Ultimately, despite arguments suggesting that the public has grown more skeptical and

cautious about the authenticity of online content, disinformation experts warned that deepfake

technology exacerbates people’s difficulty in distinguishing between genuine and forged online

content (Mozur and Satariano). A study conducted by Köbis et al. (2021) has revealed two

important insights regarding the reliability of people in detecting deepfakes. First, it shows that

the difficulty in detecting deepfakes is not due to a lack of motivation but rather a lack of ability.

Despite participants' genuine effort to identify deepfakes, they were unable to do so consistently.
 Sample A 7
This highlights the complexity and sophistication of deepfake technology, making it increasingly

challenging for individuals to distinguish between real and manipulated videos. Second, the

study reveals a systematic bias among participants toward assuming that videos are authentic,

coupled with an overconfidence in their own ability to detect deepfakes. This suggests that

people tend to rely on a "seeing-is-believing" heuristic, assuming that what they see in a video is

true. This cognitive bias puts individuals at risk of being influenced by deepfakes, as they may

trust and share manipulated content unknowingly.

In the face of advancing deepfake technology, it is evident that this technology can be

exploited to interfere with electoral processes and undermine democratic systems. Recognizing

the potential harm and the ever-evolving nature of deepfake technology is crucial in safeguarding

the integrity of electoral processes. To effectively address the risks associated with political

deepfakes, a comprehensive and multi-faceted approach is required. This approach should

encompass technological advancements aimed at detecting and mitigating deepfakes, media

literacy campaigns to educate the public about the dangers of misinformation and disinformation,

and regulatory measures to establish guidelines and accountability. Only by addressing the

challenges posed by political deepfakes head-on can we protect the foundations of our

democratic societies and secure a future where truth triumphs over deception.
 Sample A 8
Works Cited

Agence France-Presse. “Porn, politics are key targets in 'deepfakes' – study.” Rappler, 8

October 2019,

https://www.rappler.com/technology/242016-deepfake-pornography-politics-deeptrace-st

udy-october-2019/.

CBS News. “Doctored Nancy Pelosi video highlights threat of "deepfake" tech.” CBS

News, 26 May 2019,

https://www.cbsnews.com/news/doctored-nancy-pelosi-video-highlights-threat-of-deepfa

ke-tech-2019-05-25/.

Cercone, Jeff, and Madison Czopek. “Joe Biden has offered support for transgender

Americans, unlike what deepfake video shows.” PolitiFact, 6 February 2023,

https://www.politifact.com/factchecks/2023/feb/06/instagram-posts/joe-biden-has-offered

-support-transgender-american/.

Christopher, Nilesh. “Deepfakes by BJP in Indian Delhi Election Campaign.” VICE, 18

February 2020,

https://www.vice.com/en/article/jgedjb/the-first-use-of-deepfakes-in-indian-election-by-b

jp.

Esguerra, Christian. “DEEPFAKES: Ang bagong mukha ng disimpormasyon.” YouTube,

2 December 2022, https://www.youtube.com/watch?v=GDvJRI_KoFo.

Galston, William A. “Is seeing still believing? The deepfake challenge to truth in

politics.” Brookings Institution, 8 January 2020,

https://www.brookings.edu/research/is-seeing-still-believing-the-deepfake-challenge-to-tr

uth-in-politics/.
 Sample A 9
Klein, Charlotte. “‘This Will Be Dangerous in Elections’: Political Media's Next Big

Challenge Is Navigating AI Deepfakes.” Vanity Fair, 6 March 2023,

https://www.vanityfair.com/news/2023/03/ai-2024-deepfake.

Klepper, David, and Ali Swenson. “AI presents political peril for 2024 with threat to

mislead voters.” AP News, 14 May 2023,

https://apnews.com/article/artificial-intelligence-misinformation-deepfakes-2024-election

-trump-59fb51002661ac5290089060b3ae39a0.

Köbis, Nils C., et al. “Fooled twice: People cannot detect deepfakes but think they can.” vol. 24,

no. 11, 2021, pp. 9 -11. iScience, https://doi.org/10.1016/j.isci.2021.103364.

Mak, Tim, and Dina Temple-Raston. “Are Deepfake Videos A Threat? Simple Tools Still

Spread Misinformation Just Fine.” NPR, 1 October 2020,

https://www.npr.org/2020/10/01/918223033/where-are-the-deepfakes-in-this-presidential-

election.

Mozur, Paul, and Adam Satariano. “The People Onscreen Are Fake. The Disinformation

Is Real.” The New York Times, 7 February 2023,

https://www.nytimes.com/2023/02/07/technology/artificial-intelligence-training-deepfake

.html.

Ray, Andrew. “Disinformation, Deepfakes and Democracies: The Need for Legislative

Reform – UNSW Law Journal.” UNSW Law Journal, 2021,

https://www.unswlawjournal.unsw.edu.au/article/disinformation-deepfakes-and-democrac

ies-the-need-for-legislative-reform.

Reuters Staff. “Fact check: “Drunk” Nancy Pelosi video is manipulated.” Reuters, 3

August 2020,
 Sample A 10
https://www.reuters.com/article/uk-factcheck-nancypelosi-manipulated-idUSKCN24Z2B

I.

Samson, Celine Isabelle B. “VERA FILES FACT CHECK YEARENDER: As video is

king, misinformation takes its throne in TikTok era.” VERA Files, 20 December 2022,

https://verafiles.org/articles/vera-files-fact-check-yearender-as-video-is-king-misinformat

ion-takes-its-throne-in-tiktok-era.

Shyrokykh, Karina. “Fake news on social media: Whose responsibility is it?” Ericsson, 5

November 2018,

https://www.ericsson.com/en/blog/2018/11/fake-news-on-social-media-whose-responsibil

ity-is-it.

Vera Files. “VERA FILES FACT CHECK: Video of Marcos Sr. taking a jab at voters

supporting a tax evader FAKE.” VERA Files, 5 March 2022,

https://verafiles.org/articles/vera-files-fact-check-video-marcos-sr-taking-jab-voters-supp.