The risk of false political messages produced with artificial intelligence

2024-06-22 19:29:00, Kosova & Bota CNA

The risk of false political messages produced with artificial intelligence

Illustrative photo, taken from foreign media

As this year's presidential election approaches, American voters are exposed to malicious interventions that use artificial intelligence to create disinformation, either to help rival politicians or to serve the interests of foreign governments. Voice of America correspondent Ivanna Pidborska reports on the use of artificial intelligence in the 2024 elections.

Artificial intelligence has already become present in this election cycle, as in the computer-generated photographs of former President Donald Trump hugging infectious disease expert Dr. Anthony Faucin, or in the made-up voice of President Biden for a phone call asking voters in New Hampshire not to participate in the primary election.

The timing chosen for the spread of these fake materials makes them even more dangerous, says media expert Elaine Kamarck of the Brookings Institution.

"Even 24 hours before the elections, someone can put out a fake material or a disinformation. It is very difficult to fight; in very close elections this can make the difference between victory and defeat," she says.

Rijul Gupta's DeepMedia firm helps the Pentagon identify fake materials, which he says are popular because they are inexpensive and can be produced easily.

"It takes 15 minutes to create a completely fake material.

There are many such free services on the Internet. If you want to pay for something of better quality, 30 seconds of audio costs maybe 2 cents,” says Mr Gupta.

Ari Cohn is a lawyer and activist for freedom of expression.

"There's been a lot of talk about Russia, but China is also particularly involved in the artificial intelligence race to get the edge," he says.

Mr. Cohn says that the most sophisticated foreign operations aim to "plant" a certain segment of disinformation with influential Americans, so that they can then distribute it to others.

Major US tech firms and social media platforms have promised some form of artificial intelligence self-governance during election campaigns, but there are currently no federal laws on it.

“It's very, very difficult to legislate for political discourse in the United States. The First Amendment to the Constitution protects the freedom of political expression very strongly and rightfully so: We do not want the government to have a role in the outcome of the election," says Mr. Cohn.

Mr. Gupta says the biggest difficulty is differentiating between something that is false and something that describes an issue in a way that someone might disagree with.

"Poor audio quality e.g. may be an indication that it is AI-produced material. It may seem like a joke, but tracking for these small signs and quickly spotting them in images, audio and video can help to filter out a lot of them. So, if e.g. you are in a video conference with someone and you suspect that they may be fake footage, you can ask them e.g. to turn chairs. If they don't, chances are it's artificial intelligence footage," says Mr. Gupta.

Mr. Cohn says that artificial intelligence can also be used for positive purposes in politics. It could allow candidates to communicate with diaspora communities in their own language, or it could give campaign teams more accurate data on voter preferences so they can better tailor their election message. ./ Voa