SailPoint CEO: AI agents need a 'kill switch' to stop the worst from happening

Humans should be able to instantly put AI agents to sleep if they cause harm to a business.

"We talk about discover, govern, and protect. The ultimate protection is a kill switch," SailPoint (SAIL) CEO Mark McClain said on Yahoo Finance's Sozzi Unleashed. "But we're also beginning to try to get folks to understand there's a danger in an unsophisticated kill switch."

SailPoint is a key cybersecurity player sporting a market cap nearing $12 billion.

McClain noted that there are some misconceptions floating around about the concept of a kill switch. "Some of the folks that are out there kind of propagating the idea of a kill switch don't actually have very good visibility to what it is that might actually be going awry," he said.

"If there's a bad guy over there in the crowd somewhere, you don't want to just shoot into the crowd trying to get the bad guy and take out a bunch of good people," McClain added. "You need to be very clear and precise as to what exactly is bad, and make sure you don't do harm to the good things that are happening. So the concept of a kill switch is great, as long as you have a precision and a clarity of what exactly is wrong or what could be wrong, and focus only on that one bad thing."