Therobotreport iconTherobotreportSep 28, 2026 ~6 min source read

Gecko Robotics adopts NVIDIA Open Agent Safety Platform to enforce AI agent boundaries on inspection robots

Gecko Robotics is testing NVIDIA’s OpenShell runtime and Sentry watchdog to give enforceable, hardware-backed controls over agent actions such as motion, data collection, and payload handling on robots used in critical infrastructure inspections.

Gecko Robotics works with NVIDIA to add AI agent security and control

Share this story

Send the public story page.

Useful takeaways from this story.

Gecko uses NVIDIA OpenShell to create a secure runtime boundary that traces agent activity and enforces explicit policies for robot actions.

NVIDIA Sentry provides an out-of-band hardware watchdog on DPUs that continuously monitors agents and can quarantine misbehaving agents in milliseconds.

The combination of runtime policy enforcement and independent hardware monitoring aims to reduce the risk that an AI agent will bypass application-level controls and affect physical systems.

# What happened NVIDIA announced its Open Agent Safety Platform, which pairs OpenShell (an open-source runtime) with Sentry (a hardware-based watchdog). Gecko Robotics is among more than 100 organizations working with NVIDIA to test and apply these components to inspection robots that operate on critical infrastructure.

# Why this matters

# How the technology works OpenShell runs as a secure runtime boundary that records an agent's actions and enforces policies about what the agent can access and do. It can be extended to other compute platforms and can offload enforcement onto silicon or DPUs.

Sentry runs out-of-band on BlueField-4 DPUs to independently monitor agent behavior and outcomes against policy. If an agent tries to cross its allowed boundary, Sentry can quarantine that agent in milliseconds.

A Policy Proover component validates combined decision trees and access patterns to detect unintended data exfiltration or policy circumvention, using deterministic checks that evaluate file and network access together.

# How Gecko is applying the platform Gecko's inspection agents have the same system command capabilities as human field operators for specific tasks: starting and stopping data collection, controlling motion and path planning, and raising or lowering payloads. They do not conduct the data lifecycle tasks of uploading to Gecko's governance cloud.

Gecko used NVIDIA OpenShell to define what agent actions are permitted and where restrictions should be enforced. The company also places an independent enforcement layer on its Komodo robot, which has been deployed with the U.S. Navy, to add another level of control.

NVIDIA frames Open Agent Safety Platform as a full-stack approach combining software, hardware, and infrastructure to provide deterministic governance for agents. NVIDIA describes OpenShell as tracing action and enforcing policy on systems where agents run, while Sentry provides an independent watchdog.

Gecko's leadership describes safety as an engineering challenge and says the company's goal is to confront risks before they reach the physical world. Gecko emphasizes building safeguards that keep AI within the boundaries set by humans.

# What this does not change yet

# Bottom line Pairing a traceable runtime (OpenShell) with an independent, hardware-backed monitor (Sentry) gives Gecko a tooling approach to limit what AI agents can do on inspection robots. The combination aims to reduce the chance that an agent will escape application-level controls and perform unintended physical actions, while keeping agent permissions explicit and auditable.

More context around this story.

AI бЂ”бЂЉбЂєбЂёбЂ•бЂЉбЂ¬бЂЂбЂ­бЂЇ бЂЎбЂ™бЂјбЂ”бЂєбЂ†бЂЇбЂ¶бЂё бЂњбЂ±бЂ·бЂњбЂ¬бЂ”бЂЉбЂєбЂё
Medium iconMediumSep 5, 2026

AI бЂ”бЂЉбЂєбЂёбЂ•бЂЉбЂ¬бЂЂбЂ­бЂЇ бЂЎбЂ™бЂјбЂ”бЂєбЂ†бЂЇбЂ¶бЂё бЂњбЂ±бЂ·бЂњбЂ¬бЂ”бЂЉбЂєбЂё

AI (Artificial Intelligence) နည်းပညာက အá€á€¯á€¡á€á€»á€­á€”်မှာ နေရာá€á€­á€¯á€„်းမှာ ရှိနေပါပြီዠဒါပေမဲ့ “AI ကို ဘယ်ကနေ စလေ့လာရမလဲአအမြန်ဆုံး á€á€á€ºá€™á€¼á€±á€¬á€€á€ºá€¡á€±á€¬á€„်â

Loading more related stories...

Keep reading in the app

Open the app view to save this story, compare related coverage, and continue from the same source.

Open in app