New AI cyber range launches to test autonomous AI agents under real attack conditions

Hack The Box has announced HTB AI Range, a new, controlled 'AI cyber range' built to test how AI security agents behave under pressure, as well as how well they work alongside human teams.

The platform recreates real attack and defense scenarios so organizations can see how well (or if) the models cope with complex, high stakes environments. It can also flag where human oversight is still needed.

SEE ALSO: 82% of HR professionals use AI, but most lack job-specific training

AI now plays a role in many parts of enterprise infrastructure, which means the attack surface keeps growing. More autonomous tools are being used by people who might not fully understand their limitations or their risks, creating a need for controlled environments where both humans and AI agents can be evaluated before being trusted in real operations.

AI cyber range

The company’s CEO and founder, Haris Pylarinos, said: “AI is now part of the cyber battle and overall ecosystem, and we’re building the arena where it can safely be tested and used to defend responsibly. For over two years, we’ve been advancing AI-driven learning paths, labs, and research where machines and humans compete, collaborate, and co-evolve. With HTB AI Range, we’re not reacting to AI’s rise in cyber; we’re defining how defense evolves alongside it. This is how cybersecurity advances: not through fear, but through mastery.”

HTB AI Range works by mirroring enterprise scale environments with thousands of offensive and defensive targets that change over time. It benchmarks performance against frameworks including MITRE ATT and CK, NIST and NICE, and the OWASP Top 10.

Hack The Box says the new platform is for enterprises, security service providers, and government teams that want to stress test models, validate safety, and understand how well AI holds up in fast moving situations.

In an AI versus human capture the flag event earlier this year, autonomous agents solved almost all of the easy challenges, hitting a 95 percent success rate on tasks at the lowest difficulty level. They struggled with multi step problems however, where human red teams still performed far better, but the results show how attackers could use automation to scale their activity in ways that weren’t possible before.

Gerasimos Marketos, the company’s Chief Product Officer, said: “Hack The Box is where AI agents and humans learn to operate under real pressure together. We’re addressing the urgent need to continuously validate AI systems in realistic operational contexts where stakes are high and human oversight remains vital. HTB AI Range makes that possible. It’s the next step in building trust, safety and performance into AI for cyber defense.”

HTB plans to launch AI Red Teamer Certification in early 2026, which is part of a wide learning path developed with Google to help security professionals assess, test, and secure AI systems.

What do you think about AI agents being tested in live fire cyber ranges? Let us know in the comments.