As state agencies in Missouri increasingly make use of Artificial Intelligence, staff and leadership are on guard against cybersecurity threats, according to Tim Marczewski, Director of AI and Innovation with the Office of Administration.
He told Missourinet for starters, OA doesn’t use public AI tools.
“The free models that you would see, like the free ChatGPT or the Free Gemini or whatever your large language model of choice is to use, we use ones that are in a government secured tenant. So, the models are not learning off the data we’re putting in there,” he said.
Marczewski also said their AI tools have built-in security.
“Not just security slapped on at the very end. It’s built in from the beginning — also, rigorous testing and quality assurance,” he said. “So, (we’re) doing that early prototyping, doing those tests, making sure that you’re getting the answers you expect from the models.”
On top of that, he added that there will always be a human “in the loop.”
“It will never be Artificial Intelligence making decisions without a human checking that, so it will always have a human making sure that that is the correct decision,” Marczewski said. “That is the direction we want to go.”
The governor’s Office of Administration oversees more than a dozen state agencies in Missouri.
Copyright © 2025 · Missourinet