Researchers warn guardrails on ChatGPT and other chatbots easily circumvented

January 08, 2024

CBS News Pittsburgh interviews DTI researcher Zico Kolter of Carnegie Mellon University about his discovery of how easy it can be to engineer a “jailbreak” in chatbots to break through safeguards and generate harmful information.

These vulnerabilities can make it easier to allow humans to use the chatbots for all sorts of dangerous purposes, generating hate speech or fake social media accounts to spread false information — something the author fears in the upcoming presidential election, increasing divisions and making all information suspect.

“I think the biggest risk of all of this isn’t that we believe all the false information, it’s that we stop trusting information period. I think this is already happening to a degree,” Kolter said. “Used well, these can be useful, and I think a lot of people can use them and can use them effectively to improve their lives if used properly as tools.”

Read the KDKA-TV story here.

KDKA-TV/CBS News Pittsburgh graphic