Democratic elections are not safe from AI interference, a former Google boss has warned.
Eric Schmidt, who ran the tech giant for a decade, said fake images and videos of every politician would likely flood the internet during campaigning.
This spread of AI-generated misinformation would drive the public ‘crazy’ as they wouldn’t be able to work out what was true, he predicted.
Speaking to Andrew Marr on LBC, Mr Schmidt said the UK and US were ‘clearly’ ahead in the AI race – and it was vital they decided what was ‘acceptable in democracy’ when it came to regulation.
His comments come as ministers are preparing to protect the next general election from manipulation by AI over fears it could be hijacked by deepfakes spread by hostile forces.
Eric Schmidt, who ran Google for a decade, said fake images and videos of every politician would likely flood the internet during campaigning
Mr Schmidt is the second former executive from the tech giant – which has been accused of an AI arms race against its rival Microsoft – to warn of the dangers of the technology.
Earlier this month, British scientist Dr Geoffrey Hinton – known as the ‘Godfather of AI’ – quit his job at Google to warn the public about the dangers machines could soon pose humanity.
The 75-year-old said his biggest worry for the immediate future is how ‘bad actors’, such as Russian President Vladimir Putin, might use the tools to win wars and elections.
Mr Schmidt was Google chief executive between 2001 and 2011, before taking over as executive chairman and remaining in high level positions until he left in 2020.
Warning of the dangers of AI, he said he did not believe democratic elections were safe from interference from it and branded the potential it had to trick people as ‘horrific’.
He said: ‘The spread of generative AI to create false images, not just pictures, but also videos will drive the population crazy.
‘Every political character will both be the victim of all this but will also have some affiliated group that uses it. And we don’t today have a perfect way of knowing whether this was an authentic photo or not. That has to get fixed.
As cops in New York City and Washington DC prepared for possible unrest in the event of Donald Trump being indicted over allegations he paid hush money to a porn star, images emerged appearing to show the former president being arrested
‘You know, one of the problems is that as a baby, when you were born, you were taught to believe what you saw, and what you heard. And with generative AI, the majority of the images could be fake. And furthermore, the things you hear could be fake as well.’
He said the UK and US were leading the world in AI, which made it ‘important that we decide what is acceptable in a democracy.’
Prime Minister Rishi Sunak has promised the UK would be at the forefront of international attempts to put ‘guardrails’ on AI to stop it getting out of control.
Speaking at the G7 summit in Japan over the weekend, he said the UK was in a ‘natural position’, adding that it would need co-ordinated action to prevent dangerous risks from emerging.
The scariest AI developments so far: From ‘Balenciaga Pope’ to Fake Trump arrest and deepfakes
An image of Pope Francis wearing a shiny white puffer jacket, a long chain with a cross and a water bottle in his hand is the latest example of the dangers of AI.
The stylish pontiff was created by image-generator Midjourney, which was also behind the shocking fake scenes of Donald Trump being arrested by police officers in New York City.
Deepfake videos have also shown the evil powers of AI, allowing users to create clips of public figures spreading misinformation – and experts predict 90 percent of online content will be made this way by 2025.
These scary AI developments seem to be just the tip of the iceberg.
An image of Pope Francis wearing a shiny white puffer jacket, a long chain with a cross and a water bottle in his hand is the latest example of the dangers of AI
Elon Musk, Apple co-found Steve Wozniak and more than 1,000 tech leaders are calling for a pause on the ‘dangerous race’ to develop AI, which they fear poses a ‘profound risk to society and humanity’ and could have ‘catastrophic’ effects.
The AI-generated image of Pope Francis, published Friday on Reddit, made waves on the internet this week, ultimately because the public believed it was real.
‘I thought the pope’s puffer jacket was real and didn’t give it a second thought,’ tweeted model and author Chrissy Teigen. ‘No way am I surviving the future of technology.’
Experts have also weighed in on the realistic AI image.
Web culture expert Ryan Broderick said the pope image was ‘the first real mass-level AI misinformation case.’
The image, however, followed a gallery of fake photos showing what it could look like if Trump were arrested – but these were publicly known to be AI-generated.
Bellingcat journalist Eliot Higgins created the images this month, showing Trump being chased down the street by police officers while his wife Melania screams. Others show the former President in jail wearing an orange jumpsuit.
‘Legit thought these were real,’ one person tweeted, while another said: ‘We should really be putting watermarks on these that disclose they are AI-generated and not real.’
It comes after Trump claimed without evidence that he would be arrested today and called on his supporters to ‘protest, protest, protest’ in response to a possible indictment over the former president’s alleged hush money payments to porn star Stormy Daniels.
The stylish pontiff was created by image-generator Midjourney, which was also behind the shocking fake scenes of Donald Trump being arrested by police officers in New York City
Deepfake videos and images have also seen a boom online, showing influential figures relaying misinformation.
Meta CEO Mark Zuckerberg was used in a clip where he thanked Democrats for their ‘service and inaction’ on antitrust legislation.
Demand Progress Action’s advocacy group made the video, which used deepfake technology to turn an actor into Zuckerberg.
More recently, in February, several female Twitch stars discovered their images on a deepfake porn website earlier this month, where they were seen engaging in sex acts.
Currently, no laws protect humans from being generated into a digital form by AI.
