Healthcare cyber resilience in the age of AI

As cyberattacks increasingly disrupt clinical services and care delivery, healthcare cyber resilience has become a critical priority across the industry. Leaders are now treating cyber preparedness as an enterprise-wide capability that protects operations and patient safety.

In fact, nearly every healthcare organization is elevating cyber resilience as attacks accelerate in frequency and impact. But beyond immediate financial damage, these incidents also take a heavy human toll, according to Josh Howell (pictured), healthcare chief technology officer at Rubrik Inc.

“In terms of the impact to employees, their families, it’s not unusual to see a lot of staff attrition after one of these events,” Howell said. “When you can spend months getting back online and dealing with all of the fallout, nobody wants to live through that twice.”

Howell spoke with theCUBE’s John Furrier for theCUBE + NYSE Wired: MedTech Unplugged: The Future of AI in Healthcare & Life Sciences interview series, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed healthcare cyber resilience amid rising threats and the impact of artificial intelligence on healthcare operations.

Ensuring healthcare cyber resilience in the age of AI

Similar lessons keep resurfacing across multiple waves of technology and innovation, from big data to cloud. Chief among them is that information technology and security must be involved from the beginning, because security can’t be an afterthought and organizations can’t protect what they can’t see, according to Howell.

“Change control is really important, not even just on the model or the code, but also the data that you’re feeding into it,” he said. “If you’ve got a RAG corpus that changes without you knowing about it, you’re going to start getting unpredictable results. This is one of the areas we’re really interested in with Rubrik Agent Cloud.”

In practice, Rubrik Agent Cloud is about visibility and control: understanding what data AI tools touch and reducing the risk of surprises in production. That matters in healthcare, where AI often arrives through service lines or research teams before IT and security have full oversight. A key challenge is simply knowing what’s in use across the organization. Without that visibility and monitoring, an AI agent that goes wrong can cause tremendous harm at machine speed and scale, according to Howell.

“Really knowing all of the data assets that that agent can touch, what credentials or tokens that it’s using, what actions it’s taking, how are we validating that it’s still giving responses that are within what’s acceptable — all of those things are a problem, especially at scale,” Howell said.

Business management philosopher Peter Drucker once said that healthcare organizations were the most complex form of human organization, Howell noted. With that in mind, it’s important to imagine being a chief information and security officer tasked with securing and managing an array of AI systems implemented by others.

And the stakes are far from theoretical. ECRI, an organization that tests these technologies, recently stated that AI is the top technology threat to patient safety, according to Howell. Already, there are more than 500 documented instances where AI has caused patient harm, he added.

“We have deaths now that are attributable to AI going wrong. And it is a real problem of knowing what they’re doing,” Howell said. “And in some cases, you can have silent failure. Nothing has gone down.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of theCUBE + NYSE Wired: MedTech Unplugged: The Future of AI in Healthcare & Life Sciences interview series:

Photo: SiliconANGLE