Physical AI Moves From Lab to Frontline: Robotics Takes on Manufacturing Security's Convergence Problem

The line between cybersecurity and physical security has blurred, and manufacturing plants sit squarely in the crosshairs. As organizations grapple with protecting critical infrastructure, robotics and physical AI are emerging as a frontline layer in a fundamentally more proactive security strategy.

The underlying technology pairing is straightforward but powerful. As the industry framing puts it:

"Robotics and artificial intelligence are deeply intertwined technologies. Where the former serves as the "body," the latter provides the "brain," with organizations continuing to develop the technology from traditionally rigid machinery into smart, adaptive agents designed to enhance productivity and operational efficiency."

Investment is following that logic. In 2026 alone, over $23 billion has been invested in organizations focused on improving how the two technologies work together.

A warning shot from the automotive sector

The risks of cyber-physical convergence are no longer theoretical. Connected security systems can include interlinked cameras, access control and sensors that, if malicious actors attacked the central platform, could leave considerable vulnerabilities, opening assets up to theft, damage, and disruption.

One example of the impact is last year's Jaguar Land Rover incident. Stolen credentials and social engineering from allegedly state-sponsored actors gained access to the automaker's IT infrastructure, causing a major disruption. It completely halted physical production, costing the British economy an estimated $2.5 billion.

For manufacturing professionals, the lesson cuts deep: legacy systems are part of the problem. Operating older hardware and software only enables personnel to react to events rather than attempt to prevent them, which cannot suffice in a security environment that becomes more vulnerable with growing tensions in both the geopolitical and socioeconomic spheres.

A quadruped built for proactive patrol

At a 2025 event in Taipei, robot research and development company Foxlink presented its "robot dog" as a next-gen physical security asset designed to tackle the core challenges of protecting crucial assets.

The feature set implemented in the quadruped robot addressed the core physical security need: adaptive functions that facilitate a more proactive security environment. Its integrated smart sensors help observe the patrolled area, using physical AI to register, process and act on any anomalies and unusual activity. The robot incorporates NVIDIA and SYNC technology, according to Alice Guo, brand manager at Luminys, which partnered with Foxlink on the system.

The functionality enables a scalable, proactive approach to security, using the mobility of robotics and AI reasoning models to continuously learn, act on, and improve against complex security risks and threats.

Integration, not replacement

The engineering case here is not about ripping out existing infrastructure. Physical AI and robotics can integrate with legacy technologies to create a more comprehensive security system. Large language models (LLMs) can help security teams interpret high volumes of data and identify patterns across video, access-control and sensor feeds.

Deployment is already underway. Security robots are already active in various hospitals, shopping malls, and education campuses across the world. Other players are pushing inspection robotics forward as well — ANYbotics with its ANYmal quadruped, and Gecko Robotics working with NVIDIA on AI agent security and control.

The verdict from the security community is measured but clear:

"The adaptive and scalable potential of physical AI and robotics when applied responsibly can help address these concerns and keep up with the changing physical security landscape."

Applied responsibly is the operative phrase. The technology is maturing fast — the governance around it needs to keep pace.