At CES, it was all about robotics, from AI to security

Sponsored by

Your machine data knows things: Unlock it with Splunk and get game-changing insights—plus a critical resource to power AI. Tapping into machine data can boost your security and reliability. See more.

That android better be paranoid.

This year’s CES was heavy on the robots, from ice cream makers to sparring partners to more adult fare. As with any new technology, there are concerns over security, and for the devices that producers hope will see mainstream adoption, the threat increases with their connection to internal networks and systems.

The focus on robotics is part of a trend, Lenovo Global CIO Arthur Hu told IT Brew, one likely to further accelerate the integration of AI into physical technology. He believes robotics will expand the physical labor space and allow humans to do more.

“All human work, whether it’s knowledge work, whether it’s very physically based, I think that frontier of what it means to be able to do human value-added work will continue to change,” Hu said.

Drop in the AI. As we heard at CES, AI is increasingly integrated into hardware, what Hu calls “physical AI.” The AI utilized in these systems, said Akamai EVP and CTO Robert Blumofe, are deep learning models with capabilities beyond chatbots.

“A lot of the real value that we’re getting out of AI, whether it’s to detect fraud, recommend products, drive a car autonomously, or run a robot in a warehouse—in almost all cases, those are not LLMs,” Blumofe said. “Those are deep learning models that are specialized to solving the particular problem at hand.”

AI may make robotics more powerful—but beneath the surface, there’s the metaphorical iceberg.

A core issue is securing the data, said Dallas Dolen, technology principal at PwC, just like with other technologies. The difference is that, with robotics, users are potentially interacting constantly with the technology, meaning there’s more of a chance of leakage and attackers slipping through the cracks.

“People are using it because it’s fun to do new things, but it can’t just be fun—it also has to be safe,” Dolen said. “My kids get a new bike, they have got to put a helmet on.”

Secure the bag. Deral Heiland, a researcher with Rapid7 who focuses on IoT, told IT Brew that robotics’ network connections offer attackers a springboard to access internal systems. Segmentation—removing those connections—is a good start to defending against such attacks. A lack of trust has to be assumed when there are other devices on a network.

There’s no magic switch to solve the problem. But for researchers like Heiland, who have been red-teaming device insecurity for years, the problem can be identified.

“We’ve had other assessment type work where the pen-testers successfully were able to gain access to back-end networks and also access to other robotic systems,” Heiland said. “All these devices, as much as we want to think they are not independent, at a bare minimum, they have the AI engine that’s feeding control data to every one of these devices that have access to it.”

While the potential danger to robotics and other integrated devices isn’t new, it does require management. Stu Solomon, CEO of cybersecurity firm Human, told IT Brew that the convergence of infrastructure today is daunting—but luckily, we know how to protect it.

“If it’s got an IP address, it needs to be secured; if it’s going to have connectivity, it needs to be secured,” Solomon said. “These aren’t challenges that we haven’t seen before.”