A bid rigging scandal in Bellingham may involve ChatGPT

The city of Bellingham is conducting an internal investigation after evidence was uncovered that a staffer may have rigged a bidding process for a city contract to favor one vendor over another.

State and federal law usually require government entities to create a neutral bidding process to prevent vendors from rigging a bid like this. That's known as procurement fraud, and it's nothing new.

What is new, though, is that the city employee allegedly used ChatGPT to skew the outcome. At least one expert says it may be the first time AI has been at the center of this kind of scandal. KNKX reporter Nate Sanford, along with Cascade PBS, broke the story. Kim Malcom talked with Sanford about it in a recent interview.

This interview has been edited for clarity.

Kim Malcolm: How exactly did you uncover this alleged bid rigging?

Nate Sanford: Last summer, I filed a series of records requests with cities all across the state, basically just asking for records of government officials using ChatGPT. I was curious if people were using it in the first place, and also what types of things governments were using it for. And I ended up getting back thousands and thousands of pages from all sorts of different cities. It was really interesting and illuminating, but in the process of that, there was this one chat log that stood out. A city employee was working on creating an RFP, a Request for Proposals, to solicit applications for a new city utility billing software for Bellingham. And in one message, they asked ChatGPT to write contract requirements that would “exclude” this one specific company without explicitly saying so. And in another message, they asked ChatGPT to write requirements that would favor a different company, and some of the language that ChatGPT created in response to that actually made its way in.

You spoke with a lawyer who's an expert in these kinds of issues, and he looked at those messages between the staffer and ChatGPT. They stood out to him as well. What was so troubling to him about those messages?

He said he can't issue a concrete legal conclusion — and neither can I — but I think what raises a red flag about those messages is that local governments have to follow this pretty strict set of rules when they're entering into contracts and buying stuff, and the rules are there for fairness, for transparency, for neutrality; we want to make sure that taxpayers are getting the best deal possible.

There aren't any financial or personal links that have come up between the staffer and the company that have been revealed so far. Do you know what was behind the employee's decision to favor one over the other?

Yeah, it's not entirely clear, and they didn't respond to requests for comment. There were a few messages that seemed to indicate they were more just driven by a sense of urgency. They felt like they had done a bunch of research and identified the best company already through that process, and felt that going through this whole RFP process would be kind of redundant. It'd be an extra hoop to jump through. So there is that, but it is still a little unclear — and even if it was just a desire to save time, that is still something that's not necessarily aligned with the rules around public procurement.

You spoke to people at the City of Bellingham, what have they said about this?

They weren’t aware of this until I reached out, and they said they're taking it pretty seriously. The mayor says that they're basically bringing on an independent fact-finding investigator to look into this and figure out what happened, because they feel like there's still a lot of context that they don't have. They're trying to figure things out. But the mayor really stressed that they're taking it seriously. If indeed this is what it looks like, that wouldn't be acceptable.

You came across this while you were working on a series about how governments are using AI. Are there any concerns that people are bringing up to you about government workers use of AI compared to people in other sectors of the economy?

Yeah, I think so much of the debate around AI is really, really polarized, and people have pretty strong feelings about a lot of the different ways that it's used. I think in the earlier stories we published, there were a lot of different examples of people using it for researching legislation, things like asking it for policy advice, asking it for help writing speeches, writing talking points, writing press releases. And there are some examples of people using it also to respond to emails from constituents. And I ended up, you know, calling one of those constituents who received an AI-generated email response to something they'd sent to the city. And they were pretty upset when I told them that this response they’d gotten from the city was from someone who copy-pasted their email into ChatGPT and asked it to generate a sympathetic response.

Even though the response itself probably wasn't that different from what a human would have written, I think there was still kind of this feeling of inauthenticity that like, well, if they're not going to take the time to respond to me, why should I read it? And there are lots of people out there who would not have a problem with that, but I think some people also do still have strong feelings about it being used that way, especially when it's your elected officials and your representatives. It’s just a really quickly evolving thing, and everyone's still figuring out what the norms and standards should be.

Listen to the interview by clicking the play button above.