Hack The Box has introduced
HTB AI Range, a controlled cyber range where autonomous AI security agents can be tested alongside human defenders and attackers. The launch matters more as MSSPs and enterprise SOCs are under real pressure to put AI tools into production, often without a clear way to prove how those tools hold up when workloads spike or incidents get messy.
AI is now already woven into security operations, development workflows, and customer-facing systems. It brings speed and scale, but it also adds risk. Many AI-driven systems are now making decisions around detection, triage, and response before teams fully understand how those systems behave under stress. HTB AI Range is meant to close that gap, giving organizations a way to test and validate AI in realistic conditions before it becomes part of a live customer SOC.
Giving MSSPs Proof, Not Promises
For MSSPs, the stakes around AI adoption are especially high. Deploying an untested AI agent is not just a technical decision; it directly affects service quality, customer trust, and reputation.
HTB AI Range gives MSSPs a controlled, high-fidelity environment where they can benchmark, stress-test, and compare AI agents under conditions that closely mirror live SOC pressure. Instead of relying on vendor claims or limited proofs of concept, teams can observe how different agents behave when detection accuracy, response speed, and reliability actually matter.
“Instead of relying on vendor promises, MSSPs can finally see how an AI agent performs under realistic SOC pressure,” Marketos said.
That visibility matters on the customer side as well. MSSPs are increasingly expected to explain how AI is used inside their services and how risk is managed when automation is involved.
“Benchmarking provides the reassurance customers need,” Marketos said. “When MSSPs can demonstrate that every AI agent in their SOC has been rigorously tested, validated, and proven in a controlled environment, it strengthens trust and credibility. Customers welcome innovation, but they ultimately seek assurance, and transparent benchmarking delivers exactly that.”
Measuring Whether AI Can Keep Up With AI-Driven Attacks
The need for this kind of evaluation is driven by how quickly attacks are changing. Automated exploitation, AI-generated phishing, adaptive malware, and large-scale reconnaissance campaigns now operate at machine speed. Human-only SOC models struggle to keep pace.
“AI-driven attacks are evolving at a pace that human-only teams simply can’t match,” Marketos said. “Automated exploitation, AI-generated phishing, adaptive malware, and large-scale reconnaissance now operate at machine speed. To keep up, MSSPs need AI-powered systems that can detect, triage, and respond just as quickly.”
At the same time, not every AI agent is actually ready for that role.
“Not all AI agents are created equal - and deploying the wrong one into a live SOC introduces real operational and reputational risk,” Marketos added. “That’s why HTB AI Range is purpose-built to help MSSPs evaluate, benchmark, and safely operationalize AI capabilities before rolling them out across customer environments.”
HTB AI Range allows MSSPs to simulate the velocity and complexity of modern threats and directly measure how AI tools perform across tier-1 and tier-2 use cases.
“AI Range lets MSSPs determine which AI tools are capable of handling the workload typically assigned to tier-1 and tier-2 analysts, and which ones aren’t ready,” Marketos said.
This approach reflects results Hack The Box has already seen in internal testing. AI agents can perform well on simple, single-step challenges, but they still struggle with multi-stage, adaptive problems where human reasoning and contextual judgment remain essential. HTB AI Range is designed to make those limits visible before they become operational incidents.
AI Red Teaming Becomes a Core Discipline
Alongside HTB AI Range, Hack The Box has announced an upcoming AI Red Teamer Certification, expected in Q1 2026. The certification builds on the AI Red Teamer learning path developed with Google and reflects a broader shift in how MSSPs will need to approach AI security.
“MSSPs will start treating AI security as its own discipline, not an add-on,” Marketos said. “Until now, most MSSPs have addressed AI risks as an extension of cloud security, application security, or SOC monitoring. This curriculum changes that.”
The certification covers a wide range of topics, including machine learning fundamentals, data-pipeline compromise, gradient-based evasion, prompt injection, MCP exploitation, and LLM output abuse. Together, these areas reframe AI systems as a distinct attack surface that requires dedicated expertise and operational models.
“Its breadth reframes AI security as something that demands its own tooling, training paths, and accountability inside an MSSP,” Marketos said.
This shift is expected to influence how MSSPs structure onboarding, training, and managed services. AI threat modeling is likely to become a standard part of service delivery, rather than an experimental add-on.
From AI Literacy to AI Operations
Training expectations inside SOCs are also set to change. Basic familiarity with AI tools will no longer be enough. “Analyst training will move from user-level AI literacy to operator-level AI offensive proficiency,” Marketos said.
MSSP analysts may know how to use AI for alert triage today, but many lack a deep understanding of how models behave, where they fail, or how adversarial techniques exploit those weaknesses. The AI Red Teamer curriculum pushes teams to understand AI systems end to end, from data pipelines and model architectures to deployment and inference.
According to Hack The Box, tier-1 analysts will need to understand baseline AI behaviors, while tier-2 and tier-3 analysts will increasingly be expected to use AI-focused adversarial tooling alongside traditional tools like nmap or Burp.
New Roles Inside the MSSP
As AI security matures, Hack The Box expects MSSPs to build new internal specializations rather than stretching existing roles thinner. The AI Red Teamer Certification is designed to support that transition by giving organizations a common skills baseline.
Emerging roles are likely to include AI Team Operator, AI Data Pipeline Pen Tester, LLM Security Engineer, AI Evasion Specialist, AI Incident Responder, and AI Attack Surface Analyst. These positions reflect the reality that securing AI systems requires different thinking, techniques, and workflows than traditional infrastructure or application security.
Rather than framing AI as either a threat or a shortcut, Hack The Box is positioning HTB AI Range as a practical control point. It gives MSSPs and enterprises a way to test AI agents repeatedly, measure outcomes under pressure, and decide where automation ends and human oversight begins.
For security teams under pressure to adopt AI quickly, the message is straightforward: trust in AI comes from evidence. HTB AI Range is designed to provide that evidence before AI becomes part of the front line.
