
Building a Practical Evaluation for AI Agents in Cybersecurity
How Obscurity Labs built a controlled, evidence-driven evaluation for AI agents performing practical cybersecurity tasks.
Obscurity Labs
Redefining Cybersecurity & Engineering in public and private sectors
Software, AI, and cyber capabilities for intelligence, defense, and complex commercial operations.
Our capabilities
We combine applied research, engineering, and operations to solve hard problems in complex environments.
Design and build secure, resilient software systems that perform in the most demanding conditions.
Explore capabilityAdvance critical capabilities through rapid prototyping, testing, and applied research.
Explore capabilityProtect critical operations and data through offensive, defensive, and assurance capabilities.
Explore capabilityBuild the skills and readiness teams need to operate, defend, and deliver under pressure.
Explore capabilityFrom the blog

How Obscurity Labs built a controlled, evidence-driven evaluation for AI agents performing practical cybersecurity tasks.

Obscurity Labs conducted a penetration test against Rabbit Inc.'s services and R1 device, covering insights into various findings, attack paths, and security measures in place. Discover the strengths and vulnerabilities revealed through advanced testing methodologies and stay informed about the security of the Rabbit R1.

Explore the latest vulnerability in Log4j.