Imran Khan’s jail speech shows AI’s peril

During Pakistan’s electoral campaign, Imran Khan, the former prime minister, faced imprisonment and disqualification from participating in what experts deemed one of the country’s least credible general elections in its 76-year history. Despite these challenges, Khan employed a tech-savvy strategy, utilizing artificial intelligence to emulate his voice in speeches delivered from behind bars, aiming to counteract a military crackdown.

Khan’s AI-generated voice communicated with voters

In a surprising turn of events on Saturday, official counts revealed that candidates aligned with Khan’s party, Pakistan Tehreek-e-Insaf (PTI), secured the most seats, causing political upheaval. Moreover, it was Mr. Khan’s AI-generated voice that announced the victory in a minute-long video, featuring historical images and footage of Khan and including a disclaimer about its AI origin. The speech, characterized by a mellow, slightly robotic tone, expressed confidence in the voters and dismissed the victory claim of Khan’s rival, Nawaz Sharif, urging supporters to defend the triumph.

As concerns rise regarding the potential misuse of artificial intelligence, especially in the context of elections, Imran Khan’s videos serve as an example of how AI can be employed to overcome suppression. However, experts also highlight the growing apprehension about the inherent dangers associated with this technology.

Fear of manipulation using AI

Analysts are acknowledging the potential positive outcome of Khan’s use of AI to communicate with supporters while imprisoned. Still, the simultaneous erosion of trust in what people see and hear due to the technology’s ability to manipulate content.

Imran Khan, a charismatic former cricket star, faced expulsion from power in 2022 and subsequent imprisonment on charges like leaking state secrets. Khan and his supporters argue that military leaders orchestrated his removal, a claim rejected by authorities. During the election campaign, officials impeded Khan’s candidates from campaigning and censored news coverage of the party, prompting organizers to resort to online platforms for virtual rallies.

In December, Khan’s party started employing AI to disseminate his messages, creating speeches based on notes he passed to his lawyers in prison, according to party statements, and transforming them into videos.

More politicians are using AI

This use of AI in politics is not unprecedented. In South Korea, the People Power Party created an AI-powered avatar of its presidential candidate, Yoon Suk Yeol, for virtual interactions with voters. Similarly, politicians in the United States, Canada, and New Zealand have utilized AI to create dystopian images to emphasize their arguments or to showcase the technology’s potential risks through deepfake videos.

The integration of AI, particularly deepfakes, into political campaigning is not a passing trend but an evolving practice that will persist over time.