A MARTÍNEZ, HOST:
OK. Maybe at this point, you've had a little fun testing out AI chat bots, asking them to write a song, maybe you have them write email for you, but this campaign season, the company behind ChatGPT is banning politicians and lobbyists from using it for official campaign business. NPR cybersecurity correspondent Jenna McLaughlin is here. Jenna, OpenAI is behind ChatGPT. How's their ban going to work?
JENNA MCLAUGHLIN, BYLINE: Good question. So OpenAI is making this pledge to prevent official campaign workers from using ChatGPT. So, you know, presumably, if you try to sign up with a company email address, a campaign email address, that won't be possible. In reality, proving someone's actually using a chatbot can be difficult. So it might be difficult to enforce, particularly on a local level. The Federal Election Commission is trying to do their own thing. They're looking at ways to regulate AI-generated images - what's called deepfakes - in political ads. They say that hopefully by the summer before the 2024 election, they'll have done that. On the state level, some states have passed laws requiring campaigns to disclose any use of AI in advertisements.
MARTÍNEZ: Why is the company asking campaigns not to use this bot?
MCLAUGHLIN: So OpenAI is basically saying, you know, while we figure out what threats this might pose to democracy, figure out potential problems, we want you to not use it in campaigns. To be clear, it's one of several companies and this is the first to take this stand. When the tool was released, it was still in beta or test mode. It was designed with certain guardrails to prevent people from doing something bad with it, you know, like hacking. But people have found all kinds of ways around that. To use it for something as important as elections, it could pose all kinds of potential problems.
MARTÍNEZ: What kind of problems are we talking about?
MCLAUGHLIN: So there could be inaccurate or misleading information from these ChatGPT bots. Artificial intelligence like ChatGPT learns based on what information it's given, and that access is limited. It can make mistakes, it can be biased. Plus it, you know, makes it easier to craft better phishing emails, for example, I've talked to Jen Easterly. She's the director of the Department of Homeland Security cyber agency, CISA, a little bit about this. She says that AI will make it easier for bad guys to pump out better fake content more easily and for cheap.
MARTÍNEZ: And what are the campaigns saying about all this?
MCLAUGHLIN: I spoke to Stephen Boyce. He's a cybersecurity expert who advises federal and local campaigns on these kinds of issues. He's seen this issue moving really fast.
STEPHEN BOYCE: You know, in all transparency, before ChatGPT came about, there was little to no talk around AI. It was kind of like a distant and emerging threat. Once ChatGPT was released, it really reared its head in terms of the discussions, certainly, here in Washington.
MCLAUGHLIN: Now he says when he talks to campaigns, this issue comes up almost every single day.
MARTÍNEZ: So, I mean, does he have any advice?
MCLAUGHLIN: Yeah. It's a difficult problem. This will probably be one of the first major campaigns where generative AI is so convincing it seems real. It's going to take a lot of transparency to make sure voters know that information is real, where the data is coming from. One local lawmaker and Michigan rep. Penelope Tsernoglou, she played an AI-generated version of President Biden congratulating her on a new bill to combat AI-generated disinformation. Take a listen.
(SOUNDBITE OF ARCHIVED RECORDING)
AI-GENERATED VOICE: (As Joe Biden) Hi, representative Tsernoglou. It's your buddy Joe. I really liked your bill that requires disclaimers on political ads that use artificial intelligence. No more malarkey. As my dad used to say, Joey, you can't believe everything you hear. Not a joke. By the way, this statement was created using artificial intelligence.
MCLAUGHLIN: So again, not a real President...
MARTÍNEZ: (Laughter) Wow.
MCLAUGHLIN: ...Biden, but it's pretty convincing. It really shows the possibilities of how these tools can be abused.
MARTÍNEZ: That is NPR's the real Jenna McLaughlin. Jenna, thanks.
MCLAUGHLIN: Thank you.
(SOUNDBITE OF MUSIC) Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.
