The threat of AI in a year of elections

Governments and tech firms are at odds over how best to police an information ecosystem at serious risk of disruption

- Jonathan Yerushalmy

Gail Huntley recognised the gravelly voice of Joe Biden as soon as she picked up the phone one day in January. Huntley, a 73-year-old resident of New Hampshire, was planning to vote for the president in the state's upcoming primary, so she was confused that a pre-recorded message from him was urging her not to.

"It's important that you save your vote for the November election," the message said. "Voting this Tuesday only enables the Republicans in their quest to elect Donald Trump again."

Huntley quickly realised the call was fake, but assumed Biden's words had been taken out of context. She was shocked when it became clear that the recording was AI-generated. Within weeks the US had outlawed robocalls that use voices generated by AI.

The Biden deepfake was the first major test for governments, tech companies and civil society groups, who are all locked in heated debate over how best to police an information ecosystem in which anyone can create photo-realistic images of candidates, or replicate their voices with frightening accuracy.

Citizens of dozens of countries including the US, India and most likely the UK - will go to the polls in 2024, and experts say the democratic process is at serious risk of being disrupted by artificial intelligence.

Watchdogs are warning that with more than 40,000 layoffs at the tech companies that host and moderate much of this content, digital media is uniquely vulnerable to exploitation.

Mission impossible? 

For Biden, concerns about the potential dangerous uses of AI were expedited after he watched the latest Mission Impossible movie. Over a weekend at Camp David, the president viewed the film, which sees Tom Cruise's Ethan Hunt face down a rogue AI.