WASHINGTON — Access to artificial intelligence is skyrocketing. And so is the danger AI could pose to US elections. Technology experts say it is a “no brainer” AI will be used to try and influence our elections.
In fact, it already is.
Just hours after President Joe Biden announced his reelection bid on April 25, the Republican National Committee released this video warning four more years would be bad for the country.
But the 30-second ad filled with images of war and despair wasn’t real. The RNC admitted the ad was built entirely of images created using artificial intelligence. Along with fake reports of what sound like news reporters.
The RNC included a small watermark in the corner of the video disclosing it was “built entirely with AI imagery.” But with tools to create fake text, audio or video now readily available to anyone on the internet, often for free or little money, companies that track AI believe it will be increasingly misused to influence US elections.
“There is no doubt in my mind,” said Bryan Ware, the former assistant director of cybersecurity in the U-S government’s Cybersecurity and Infrastructure Security Agency (CISA.) Ware is also the Chief Development Officer of the cybersecurity firm ZeroFox.
Ware said AI can be used “in the hands of an adversary that wants to sow discord that wants to cause doubt in the veracity of the elections or wants to push a message that might be a fringe message but give it amplification.”
“Those are the areas where I think we'll see these AI tools," he said.
Ware explained that AI has already influenced international politics including Turkey’s most recent election, where manipulated video footage was allegedly used by Turkey’s President to influence public sentiment against his main opposition candidate.
It doesn’t even take perfect fakes to mislead people. In fact, cybersecurity firm Check Point says there can be noticeable signs an image is computer generated. AI has particular trouble re-creating hands. But the more data someone has to create the AI image or video, the better and the more realistic it will appear.
“So a public figure, like Trump or Biden, where there is you know, millions of hours of video and audio of these individuals, that can be used to train the AI,” said Tony Sabaj, Check Point’s Chief Technical Officer.
Unfortunately, when asked about guardrails to limit the misuse of AI in our political system, Sabaj said there aren’t many right now.
Experts say journalists play a huge role, with segments like WUSA9 and Parent Company TENGA’s “VERIFY” being an invaluable way of flagging and debunking political A-I fakes.
Because they say at least for now, AI is a lot easier to use than it is to counter.
“I think the good news behind that though, is that we are all going to encounter it,” Ware said. “But it will prepare us for what’s likely to happen in this next election.”
The Brookings Institute says AI can help political adversaries more effectively target the small number of undecided voters who often swing an election...or even spread phony messaging aimed at keeping large voting blocs at home on election day.
Social media companies are trying to crack down on the misuse of AI.
The US government continues to investigate new regulations and potential penalties for misusing AI to influence elections.