Enterprises need to think of agents as ‘digital co-workers’ – and that means implementing the same security safeguards

Agents are weaving their way into frontline operations across a range of industries, according to Cisco’s chief product officer Jeetu Patel, and that means security is more important than ever.

Speaking during the opening keynote session at RSAC Conference 2026, Patel told attendees that agents should now be thought of as “digital co-workers” in enterprise settings.

Indeed, human workers are operating in tandem with agents, with the latter carrying out tasks on their behalf to help alleviate strain and reduce manual toil.

While this offers huge advantages for workers, there are some drawbacks, Patel noted, particularly given they require deep access to sensitive enterprise data. It’s here that risks are elevated for those using agents, he added, meaning safeguards are key.

“You have to give them access to all your systems so that they can work on your behalf,” he said.

“Here’s the fact with these digital co-workers: you can’t really go check their resumes. You can’t make reference calls to make sure that these agents are working well, you can’t do any background checks.”

Key concerns at present rest on the fact agents “follow instructions very literally”, meaning there’s room for potential mistakes.

“They operate without any fear of consequence,” he said. “I mean, what are you going to do, fire an agent? What you’re starting to see is there’s a tremendous amount of complexity.”

Given the risks involved, Patel said a critical imperative for security teams moving forward will be to “make sure that we protect the world from agents that might go rogue”.

But security teams operate at human speed, he noted, meaning they need to essentially fight fire with fire, leveraging AI-based tools to monitor agent activities.

This is an area in which Cisco has been building tools aimed specifically at improving safety, observability, and security. Among these is the launch of Defense Claw, a security framework for Open Claw deployments.

“That means if you’re using Open Claw and you want to make sure you’re safe and secure,” he said. “This is completely open sourced.”

Treating agents like humans

Patel noted that long-standing practices such as zero trust are applicable in helping monitor agent activities. Simply put, agents should be treated with the same level of scrutiny that humans have in recent years.

“For years we’ve talked about this notion of zero trust to provide what we call least privileged access to humans, which means only give it permissions to do things that you want it to do,” he explained.

With agents, Patel said enterprises will need to do “much more than just access control”. Indeed, given agents act autonomously, there needs to be a shift away from access control toward “action control”.

“Which is going to need to verify behavior and provide control to the agent when the agents start going awry.”

FOLLOW US ON SOCIAL MEDIA

Follow ITPro on Google News and add us as a preferred source to keep tabs on all our latest news, analysis, views, and reviews.

You can also follow ITPro on LinkedIn, X, Facebook, and BlueSky.