28 Zero-day results in less than two days: what the discovery of the speed of artificial intelligence means for robotics

VicOne identified an initial set of 28 zero-day results in less than two days using AI-assisted research on software and tools that support the physical AI ecosystem. These include projects developed or maintained by Google, NVIDIA, and Meta, among others.

Based on initial validation, all 28 findings are classified as zero-day vulnerabilities. The count may change due to vendor and vulnerability authority review, and some results do not yet have CVE tracking numbers.

Many of these tools help teams train, simulate, test, and operate robots. Their weaknesses could affect the models teams build, the test results they trust, and the instructions sent to the robots.

These findings create two demands for robotics teams: respond as AI helps uncover weaknesses more quickly, and evaluate how each could affect robot safety and operations.

What artificial intelligence changes in the discovery of vulnerabilities

Discovering vulnerabilities involves examining code, tracking how systems communicate, and testing whether a suspected weakness can actually be exploited. AI can speed up parts of this work, helping researchers analyze more software in less time. Attackers could use similar functionality.

When weakness is a zero-day: Previously unknown and without a fix available — Faster discovery may leave teams little time to respond. While such exposure does not necessarily mean an attack has occurred, teams still need to determine whether they are using affected software and what an attacker could access or modify.

Where the results fit into the robot’s life cycle

VicOne results span three phases of the robotics lifecycle: development and training, simulation and validation, implementation, and runtime. Weaknesses at each stage could affect what a robot learns, how it is tested, and how it is controlled.

The following scenarios show how a compromised component could affect the robot at each stage.

• Development and training: Hidden changes can reach the finished robot
  A team downloads a pre-trained model to adapt to their robot. Uploading a malicious file could give an attacker access to the training system, allowing changes to the training data or the model itself. A manipulated model might behave normally during routine testing but respond incorrectly to a particular object, scene, or command after implementation. The risk can enter through a tool that the team trusts, before anyone attacks the robot directly.
• Simulation and validation: a passed test could give false confidence
  Importing a compromised scene or robot model could allow an attacker to gain access to the simulation environment. With sufficient control, the attacker could alter simulated sensor readings, physical conditions, or test results. A robot might then appear to pass tests that should have revealed unsafe behavior, leading the team to approve it for real-world use based on misleading evidence.
• Deployment and runtime: Operators may lose visibility or control
  An attacker could exploit weaknesses to modify files that affect motion planning or send malicious map data that blocks the operator’s monitoring display. Compromising a computer operating remotely could also jeopardize instructions sent to the robot. Depending on the system and its protections, these weaknesses could alter the robot’s movement or leave operators unable to see what is happening and respond promptly.

These are potential consequences and not demonstrated outcomes for each outcome. The actual effects depend on attacker access, system design, and whether other safeguards prevent the changes from reaching physical behavior.

In a separate demonstration, VicOne researchers provide a concrete example of a weakness affecting a protective function. Researchers exploited a vulnerability to prevent a robot’s remote stop command from working. The case shows why teams need to test whether security measures remain effective when the software supporting them is compromised.

A published patch has yet to be proven safe

AI can shorten the search for weaknesses, but a robotics team still needs to determine whether a solution is safe to implement. Even after a software maintainer releases a patch, teams must identify affected systems, verify compatibility, and verify that the update maintains expected behavior.

A previous successful test remains useful evidence. However, a newly discovered weakness may reveal conditions that the test never examined.

Three questions can help teams answer:

• Where do we use the affected software? Include training computers, simulation environments, and distributed robots.
• What could change the weakness? Consider whether it could affect models, test results, operator visibility, or robot behavior.
• How do we verify protection? Confirm that the fix resolves the weakness and that the updated system still works as expected.

When immediately upgrading is not practical, teams may need temporary safeguards, such as limiting access to an affected service, while a validated fix is ​​prepared.

From discovery to more secure implementation

As details of the 28 findings become available through coordinated disclosure, the next step for robotics manufacturers will be to link each finding to its potential effect on the development, implementation and behavior of robots.

This work also requires understanding how threats might affect robot behavior and how to protect systems already in operation. VicOne Radeis helps teams identify potential zero-day vulnerabilities, evaluate the AI ​​model and system security, and use simulation to evaluate how these risks might affect robot behavior before deployment. VicOne Rthena supports threat detection and responses within approved policies during operation.

Faster vulnerability discovery is important when teams can turn findings into actions: identify what needs attention, verify that safeguards are still valid, and confirm that a fix does not introduce new risks. This is how cybersecurity strengthens the safety of robots as systems continue to change.

Content sponsored by VicOne