AI experts and tech-savvy political scientists are sounding the alarm about the unpredictable nature of AI generation tools like ChatGPT ahead of the 2024 US presidential election.
"The US is not prepared for that."
Fox News reported on May 15th that AJ Nash, Vice President of AI at ZeroFox, a cybersecurity company based in Minnesota, warned Americans that they are unprepared for the prospect of dealing with generative AI. According to him, generative AI could not only quickly create emails, text messages, and video clips for election campaigns, but could also be turned into dangerous tools.
Governments around the world are racing to find ways to regulate AI tools.
"Generative AI brings about a significant leap forward in the ability to create audio and video content. When you can do that on a large scale and distribute it across social media platforms, the technology will have an enormous impact," Nash explained. Among the many capabilities that generative AI like ChatGPT can achieve, there are quite a few that could strongly influence elections and voting in a modern country like the United States.
ChatGPT can generate a lot of content in a short amount of time.
Illustration photo: REUTERS
For example, ChatGPT can generate automated voice messages mimicking candidates' voices, instructing voters to vote on the wrong day; release fake audio recordings of candidates confessing to crimes or creating racist content; and create video clips of someone giving a speech or interview that never actually happened. In addition, computer-generated images based on local news reports can appear, containing misleading information such as announcing a candidate's withdrawal from the race.
"What would happen if billionaire Elon Musk personally called and persuaded you to vote for a specific candidate?" asked Dr. Oren Etzioni, former founding CEO of the Allen Institute for AI (based in Washington state, USA). "Many people would listen to him. But he didn't make that phone call; it was AI," Dr. Etzioni pointed out.
Threats from the opposing side.
Petko Stoyanov, Global Chief Technology Officer of cybersecurity firm Forcepoint (based in Texas, USA), warned that foreign powers could exploit AI to try to disrupt the 2024 US election. "What happens if a foreign power, a cybercrime organization, or a state tries to impersonate someone? What would the impact be? Do we have a response?" Stoyanov posed these questions. According to him, the US will see a wave of disinformation from foreign sources during the election period.
Senator Pete Ricketts (Nebraska) also warned of the risk of pro-China organizations using deepfakes (technology that creates fabricated images and videos to look and sound like real people and events) during the 2024 election season. "They absolutely have this capability, and that's something we need to be really wary of," Ricketts said in an interview with Fox News.
Even since the beginning of the year, AI-generated fake news has repeatedly caused a stir on social media. Examples include a fake video clip of Biden making attacks on transgender people, or fabricated images showing children studying Satanism in libraries. Many people even blindly believed the internet after photos of former President Donald Trump in police records appeared, even though he was never photographed committing a crime. Other AI-generated images showed Trump resisting arrest, and of course, those were also fake.
Representative Yvette Clarke (New York) introduced a bill requiring candidates to notify voters about AI-generated advertising images. Clarke also co-sponsored a bill requiring anyone creating composite images to add a watermark clearly indicating the image's source. Several states have also introduced their own proposals to address concerns about deepfakes.
Senator Ricketts, on the other hand, suggested that what can be done now is to increase education among Americans about the dangers posed by AI-generated images. He argued that the US government could collaborate with universities to improve their ability to detect fake images before the election begins.
ChatGPT has been upgraded.
According to Mashable, OpenAI (California, USA) has just made a significant upgrade to ChatGPT, allowing the tool to connect and browse the web while running. This enables ChatGPT to update with the latest information and provide more accurate answers. Currently, corporations like Amazon are hiring AI engineers to build and integrate interfaces like ChatGPT for customer communication.
Source link
Comment (0)