Written by 1:47 am AI, Deepfakes, Discussions, Latest news

– Georgia’s Initiative to Combat AI Deepfake Fraud in Social Circles

Bill working its way through state legislature seeks to stop use of AI tech in politics – but some …

When dealing with policy matters, it may be beneficial to vocalize a problem directly in front of you.

In Georgia, the name that resonates is that of the state legislator Colton Moore. However, the representation is solely in the voice of Colton Moore.

Todd Jones, a Democratic state representative who sits on the Georgia House committee focusing on technology and infrastructure progress, has introduced a bill aiming to prohibit the utilization of deepfakes in social discourse. To emphasize his point, Jones exhibited an algorithm-generated image to the committee, depicting an AI version of Moore alongside Mallory Staples, a former Democratic congressional candidate who currently leads a far-right activist organization known as the Georgia Freedom Conference.

The video employs an AI tool to make it appear as though Moore and Mallory are endorsing the passage of the bill, with a constant disclaimer at the video’s bottom referencing the bill’s content.

Both Moore and Mallory are in opposition to the proposed regulations.

“Why should the utilization of my biometric data, such as my voice and likeness, be used to promote a policy that clearly contradicts my First Amendment rights,” questioned the AI representation of Moore.

The video proceeds, highlighting, “The majority of Georgians view the unauthorized use of my personal attributes as deceitful, yet our current laws do not address this issue. Anyone can be portrayed as endorsing a particular piece of legislation through the use of AI, even if they have never expressed such views.”

The video was produced by Brad Thomas, a Republican who co-authored and supported the bill, utilizing readily available tools.

“The specific tool we employed costs around \(50. With a \)1,000 version, even your own mother wouldn’t be able to discern the disparity,” Thomas remarked.

Thomas noted that the advancement of AI technology for creating lifelike visuals far surpasses the existing legislation required to prevent potential abuses. “These individuals appear entirely authentic, yet they are entirely AI-generated.”

The bill was approved by the committee with an 8-1 vote.

Moore is not a stranger to Georgia’s political circles. In September, his Republican peers in the state senate accused him of disseminating false information about another conservative while campaigning unsuccessfully for a program to remove Fulton County counsel Fani Willis from office.

Following a contentious speech at a memorial service held on the house floor, Moore was subsequently banned from the Georgia house chamber both verbally and physically.

Moore declined to provide a statement through the Georgia Republican media outlet.

Expressing his dissent on social media, Moore criticized the bill as an attack on the use of “memes” in social discourse, asserting that humor is a form of protected expression.

Staples, in communications with her followers, referenced the federal conviction of Douglass Mackey last year as a cautionary tale. In November 2016, Mackey, also known as the alt-right influencer “Rickey Vaughn,” disseminated numerous messages encouraging Black voters to “vote by text” rather than casting actual ballots, purportedly on behalf of the Clinton campaign.

Federal courts dismissed Mackey’s First Amendment arguments, ruling that his communications constituted fraudulent activities not shielded by constitutional rights. Mackey was sentenced to seven months in prison in October.

House Bill 986 establishes penalties of two to five years in prison and fines up to $50,000 for fraudulent election interference and solicitation of fraudulent election interference.

Under this bill, individuals who distribute, broadcast, stream, or upload significantly deceptive media within 90 days of an election will be held accountable if the content substantially impacts a candidate’s or referendum’s electoral prospects or disrupts the election process. Consequently, the use of deepfakes to challenge the outcome of an election will be deemed unlawful.

The introduction of deepfakes into the 2024 election cycle was marked by an AI-generated voice message purportedly from Joe Biden advising New Hampshire residents against voting. Subsequently, the Federal Communications Commission prohibited the use of AI-generated voice robocalls. Advocacy groups have been urging the Federal Elections Commission to establish guidelines regarding the use of AI in social advertisements. Legislation is trailing behind the capabilities of AI in deceiving the public.

States have taken the initiative by proposing and, in some instances, passing bills mandating disclosures on political advertisements utilizing AI, in the absence of federal regulations governing AI-generated content in elections. Failure to comply with these disclosure requirements renders AI-generated political content illegal in the majority of state-proposed bills.

Experts caution that AI-generated audio has the potential to mislead voters since listeners lack contextual cues that could indicate the fraudulent nature of a video. The creation of voice deepfakes depicting prominent figures like Trump and Biden is uncomplicated and cost-effective with widely available applications. These models can be further trained using ample samples of lesser-known public speakers’ voices, such as speeches or media appearances.

Enforcing the Georgia law poses a significant challenge. Prior to the emergence of AI technology, lawmakers grappled with curtailing the dissemination of misinformation and fraud through private flyers and robocalls.

“We have empowered the attorney general’s office for this reason,” Thomas explained. “Another measure we’ve taken is granting the Georgia Bureau of Investigation the authority to probe election-related issues. By leveraging the capabilities of these two agencies, we enhance our ability to identify the perpetrators.”

Legislators are only beginning to comprehend the implications of AI technology. Thomas anticipates the enactment of additional regulations in the near future.

“Fraud is fraud, and that’s precisely what this bill aims to address,” Thomas emphasized. “There are no exceptions for anyone.”

Rachel Leingang contributed to this report.

Visited 5 times, 1 visit(s) today
Tags: , , , Last modified: March 21, 2024
Close Search Window