GRAND RAPIDS, Mich. (WOOD) — A Grand Rapids-based software developer said his system flagged a racist text message before the user was able to send it out.
Lance Beaudry, the CEO and co-founder of mass messaging software TextSpot, contacted News 8 after seeing a story about racist text messages targeting Black people across the country following the election.
“We got a flag in our system about 4 o’clock on November 6 and it was the message that, come to find out, about 24 hours later, a lot of people around the country were getting,” Beaudry said.
TextSpot was originally a “side project” for Beaudry, who helped design it to be a tool for small to medium-sized businesses to send reminders, event information or other important details to a large group of recipients. It has grown to serve hundreds of customers over the years.
Unfortunately, people with ill intentions try to use it.
“A lot of spam, a lot of phishing out there in the text messaging space,” Beaudry explained. “Years ago, we realized that people were coming to our platform trying to use it for phishing and spam, so it was probably about a year and a half ago really when AI really took off, ChatGPT got huge, we were like we need a way to make sure that messages being sent from our system are not going to telecom networks being delivered to people that shouldn’t be.”
The TextSpot team utilized ChatGPT to develop an AI system that would review every message for brand-new customers. It looks for terms associated with sex, hate, alcohol, firearms, and tobacco and prohibits users from sending messages related to any of those topics.
It’s that technology that Beaudry credits for preventing a racist message from being sent out the day after the presidential election.
“It’s shocking. It’s like we disagree with it to the highest extent, right? It’s awful and we see this technology as being something that we want to encourage positive messaging, right? So we’re incredibly happy that our system did its job and did not allow this,” Beaudry said. “It caught it and it even gave it the highest severity, which was great.”
They banned the user, who signed up with the likely fake name Amy Jones, and didn’t think much of it until they realized messages similar to the one TextSpot blocked were still being sent out.
“We were like ‘Oh my gosh, this is a bigger deal,'” Beaudry recalled. “That means that other providers like us that have software like us allowed these messages to be sent, but we didn’t. We stopped it before it could spread any further.”
TextSpot users are required to verify their account with a phone number, which Beaudry said that user did. News 8 called the number, but it went straight to voicemail.
“Knowing the intentionality of what somebody was trying to accomplish here, I’m sure they used a number that couldn’t easily be traced back to an actual name to identify the person,” Beaudry explained.
The TextSpot team was able to identify an IP address that was traced to Philadelphia. Since it was under a Comcast IP address, Beaudry said it’s highly unlikely the person was using a VPN, which would have allowed them to fake the location they were signing up from.
“We can’t say with 100% certainty, but it is very likely that this was domestic. It was somebody located in Philadelphia,” Beaudry said.
Beaudry reported all the information about the user to Grand Rapids police and the FBI. He added that his team plans to cooperate with investigators by providing whatever information they need to find whoever is responsible for these messages.
Several federal and state agencies, including the FBI, FCC and Justice Department continue to investigate who is behind the messages, which showcase just how vulnerable some technology systems can be.
“We are even being more cautious now and going to probably work on just even like stress testing and seeing what are other methods that people could use to potentially get around our system? Because we don’t want that. We’re trying to create a solution, a tool that is good for people and doesn’t allow hate to spread,” said Beaudry. “I would encourage other people that are kind of that middleman, that medium between messaging and content, to consider using AI for this very purpose.”