Imran Khan, Pakistan’s former prime minister, has spent the duration of the country’s electoral campaign in jail, disqualified from running in what experts have described as one of the least credible general elections in the country’s 76-year history.
But from behind bars, he has been rallying his supporters in recent months with speeches that use artificial intelligence to replicate his voice, part of a tech-savvy strategy his party deployed to circumvent a crackdown by the military.
And on Saturday, as official counts showed candidates aligned with his party, Pakistan Tehreek-e-Insaf, or P.T.I., winning the most seats in a surprise result that threw the country’s political system into chaos, it was Mr. Khan’s A.I. voice that declared victory.
“I had full confidence that you would all come out to vote. You fulfilled my faith in you, and your massive turnout has stunned everybody,” the mellow, slightly robotic voice said in the minute-long video, which used historical images and footage of Mr. Khan and bore a disclaimer about its A.I. origins. The speech rejected the victory claim of Mr. Khan’s rival, Nawaz Sharif, and urged supporters to defend the win.
As concerns grow about the use of artificial intelligence and its power to mislead, particularly in elections, Mr. Khan’s videos offer an example of how A.I. can work to circumvent suppression. But, experts say, they also increase fear about its potential dangers.
“In this case, it’s for a good end, perhaps an end we’d support — someone who’s locked up on trumped-up charges of corruption being able to speak to his supporters,” said Toby Walsh, author of “Faking It: Artificial Intelligence in a Human World” and a professor at the University of New South Wales. “But at the same time, it’s undermining our belief in the things we see and hear.”
Mr. Khan, a charismatic former cricket star, was ousted from power in 2022 and jailed last year, accused of leaking state secrets among other charges. He and his supporters have said military leaders orchestrated his removal, an accusation they reject.
During the election campaign, officials prevented his candidates from campaigning and censored news coverage of the party. In response, organizers held online rallies on platforms like YouTube and TikTok.
In December, his party began using A.I. to disseminate Mr. Khan’s message, creating the speeches based on notes he passed to his lawyers from prison, according to statements from the party, and putting them into video.
This is not the first time political parties have used artificial intelligence.
In South Korea, the then-opposition People Power Party created an A.I. powered avatar of its presidential candidate, Yoon Suk Yeol, which interacted virtually with voters and spoke in slang and quips to appeal to a younger demographic ahead of the 2022 vote. (He won.)
In the United States, Canada and New Zealand, politicians have used A.I. to create dystopian images to drive home their arguments, or to reveal the technology’s potentially dangerous capabilities, as in a video with Jordan Peele and a deepfake Barack Obama.
During the 2020 state election in Delhi, India, Manoj Tiwari, a candidate from the ruling Bharatiya Janata Party, created an A.I. deepfake of himself speaking the Haryanvi dialect to target voters in that demographic. Unlike the Khan video, it did not appear to be clearly labeled as A.I.
“The integration of A.I., particularly deepfakes, into political campaigning is not a passing trend but a trend that will continue to evolve over time,” said Saifuddin Ahmed, an assistant professor in at the school of communication and information at Nanyang Technological University in Singapore.