Hackers Are Learning to Trick AI. Your Accounts Could Be Next

Today KSL’s Maria Shilaos talks with KSL consumer investigative reporter Matt Gephardt about a growing AI security concern: hackers trying to manipulate the bots companies use to verify identities, reset passwords, and protect accounts. It’s called prompt injection, and the worry is not just that AI can trick people. It’s that people can trick AI. Matt explains how bad actors may be able to talk an AI “gatekeeper” into bending the rules, why reused passwords can turn one compromised account into a bigger problem, and the simple protections that still matter: strong unique passwords, two-factor authentication, tighter privacy settings, and being more careful about what you post online. Plus, fire crews battle a fast-moving wildfire near Beaver as smoke blankets the Wasatch Front, doctors warn about the health effects of wildfire smoke, the University of Utah announces a new health care center at Point of the Mountain, and Before You Go takes us to This Is The Place Heritage Park, where hundreds of lanterns will light up the water in a night of remembrance, gratitude, and reflection.