The Dawn of AI-Generated Disinformation in Elections: A New Challenge for Democracy
In the lead-up to this year's presidential election, the widespread availability of generative AI — a groundbreaking form of artificial intelligence that crafts new content, including images, audio, and video — poses unprecedented challenges. There's a growing concern about the potential deception of millions of voters through the influx of political deepfakes. Despite the magnitude of this issue, Congress has barely moved the needle in formulating a response, leaving states to take the lead in crafting measures against AI-generated disinformation, albeit the effectiveness of such measures remains in question.
A poignant example of the danger was last year's incident involving a fake AI-generated audio recording, which depicted a conversation between a Slovakian politician and a journalist discussing plans to rig the country's election. This event served as a global wake-up call, highlighting the threat to democracies worldwide.
The situation in the United States further underlines the urgency of addressing this AI threat. In an alarming event preceding the New Hampshire primary, voters received a robocall with an AI-generated voice impersonating President Joe Biden, which falsely urged them not to vote. Following this incident, the Federal Communications Commission took a stand by restricting robocalls involving AI-generated voices.
Adding to the complexity, conservative groups are reportedly utilizing AI tools to facilitate mass voter registration challenges, stoking fears that this technology could amplify voter suppression efforts. "Instead of voters looking to trusted sources of information about elections, AI-generated content can grab the voters’ attention," noted Megan Bellamy, vice president for law and policy at the Voting Rights Lab.
At a time when the dissemination of "ordinary" online disinformation already threatens electoral integrity and with disinformation efforts seemingly unchecked, states are stepping up to confront the AI menace. Since last year, 101 bills addressing AI and election disinformation have been introduced. Among these, Oregon has recently joined other states like Wisconsin, New Mexico, Indiana, and Utah in enacting laws specifically targeting AI-generated election disinformation.
A new report by the NewDEAL Forum provides a blueprint for combating the AI threat effectively. Recommendations include mandatory labeling for AI-generated campaign material, conducting role-playing exercises, establishing rapid-response systems, and public education initiatives.
However, despite these efforts, several proposed measures appear hamstrung by weaknesses and loopholes, potentially limiting their effectiveness in safeguarding voters against AI. For instance, many bills mandate disclaimers for AI-generated content but do not address other forms of synthetic content such as Photoshop and CGI — dubbed "cheap fakes" — which can be equally misleading. Furthermore, some laws lack criminal penalties, raising concerns over their deterrent effect.
Given the novelty of the AI threat, states are navigating uncharted waters, striving to anticipate its manifestations. "They’re just not sure what direction this is coming from but feeling the need to do something," Bellamy observed, emphasizing the ongoing evolution of both the threat and the responses to it.
In summary, as artificial intelligence continues to evolve, the race to protect democracy from AI-generated disinformation is on. States are pioneering efforts to address this modern challenge, yet the quest for comprehensive and foolproof solutions goes on.
Analyst comment
Positive news: States are taking the lead in addressing AI-generated disinformation in elections and have introduced bills to combat the issue. A new report provides recommendations for effective strategies. The Federal Communications Commission has restricted AI-generated robocalls.
Short analyst comment: States are actively addressing the AI threat in elections through legislation and actions. However, concerns remain about weaknesses in proposed measures and the ongoing evolution of the threat.
