Generative AI is revolutionary in many ways, transforming nearly every sector and industry. But that doesn’t mean it’s not without risks. Its capacity to generate text, images, and even videos is nothing short of extraordinary; however, these same abilities mean that these new tools could be used to create fake and false information. Certainly, promoting misinformation and disinformation is concerning at all levels, but it’s especially problematic because of the various elections that are scheduled to take place this year. The impact of AI on elections is becoming a worry for many countries–including the U.S.–and it’s this AI threat to U.S. elections that demands solutions.
In total, there will be 60 different countries that will be holding elections in 2024. That means roughly half of the world’s population will be going to polling stations to cast their vote. Given this, the impact of AI on elections could be tremendous, particularly given its ability to reach broad audiences. In the U.S., evidence that this is already occurring exists with OpenAI most recently reporting the latest attempt. While none of these AI-generated disinformation campaigns have had any significant effect, the AI threat to U.S. elections is real. And its potential to be highly disruptive in this regard extends well beyond voter influence alone. As this threat grows, the ability to detect and deter fake AI-generated content is desperately needed. The stakes are simply too high to let this go unchecked.
The Potential for AI Election Influence
In past U.S. elections, fake reports, articles, and most importantly, social media posts were created as part of disinformation campaigns. These have had significant influence in influencing voter opinions and perceptions according to many reports. But this was before the era of generative AI. Generative AI offers additional tools to aid those wanting to create fake posts and disinformation. In addition to its ability to generate texts of various kinds, it can also create images and videos to coincide with these reports. Thus, the impact of AI on elections increases the risks of disinformation greatly. Without the ability to determine what’s AI-generated or not, voters are highly vulnerable to AI’s threat to U.S. elections.
Understanding this, most people appreciate deep fake content involving presidential candidates may exist. For this reason, they are often more skeptical of unusual claims. But when it comes to lower-ranking election officials and even polling supervisors, their guard may not be up. Given this, the impact of AI on elections extends well beyond major candidates. Likewise, the AI threat to U.S. elections may also include information about where to vote or even whether to vote. Previous disinformation campaigns have had a major candidate encouraging voters not to turn out. And some campaigns target the most marginalized in society to undermine their election impact. All of these actions serve to undermine trust in election integrity and in candidate information itself. With generative AI, these concerns have escalated greatly in the last two years.
OpenAI’s Latest AI-Threat Report
Unfortunately, the ability to detect the AI threat to U.S. elections is limited. To assist with these efforts, some Big Tech companies are trying to do their part. Meta has made reports of AI disinformation campaigns previously. Likewise, in May of this year, OpenAI reported that it had identified five different AI-generated AI disinformation efforts. These campaigns originated from a variety of foreign actors including Russia, China, Iran, and Israel. Notably, there was worry about the impact of AI on elections from these efforts. But in each case, the reach of these campaigns were insignificant with no appreciable influence. Regardless, it’s evident such events are occurring regularly.
The most recent AI-generated disinformation effort was just reported this past month, an AI threat to U.S. elections that originated from an Iranian source. OpenAI was able to disrupt the campaign without it having any effect. Reportedly, this was part of a larger Iranian program known as the Storm 2035 campaign. The misinformation propagated involved a variety of hot topics including the Gaza war between Israel and the Hamas. But interestingly, some content was progressive while other posts were conservative. In any case, the potential of an impact if AI on elections in the U.S. is clear. And relying on Big Tech to detect each occurrence as a solution is certainly less than ideal.
Exploring Bold Solutions to AI Disinformation
When it comes to opposing the AI threat to U.S. elections, surveillance by OpenAI, Meta and others is important. At the same time, federal agencies and Congressional legislatures are proposing a number of laws and regulations to address it as well. But in both cases, there are limitations. U.S. laws and regulations will do little to deter foreign adversaries. As such, the impact of AI on elections will remain significant. Other strategies involve educating the public about the capabilities of AI to produce high-quality content. Demonstrating its potential for generating misinformation will also help, but again, fail to truly prevent election influence. Each of these strategies should be part of a more comprehensive approach, but these alone will be insufficient.
From one perspective, Big Tech working with AI systems could do more than they are. Rather than waiting for regulatory oversight to dictate requirements, companies like OpenAI could make the first move. For example, investing in a digital labeling system that identifies content as being created by AI would be a great start. At present, such a system is lacking that could effectively target AI-generated content. Though it may have some downsides, it nonetheless could significantly reduce the impact of AI on elections. At the same time, better copyright protections would enable AI content to better reveal its sources. This could then be used to support credibility of information, also reducing the AI threat to U.S. elections. These are the types of solutions we need to safeguard future American elections of all types.
Outsourcing is essential for business growth–read why in this Bold story.
