Hack The Box has launched AI Range Enterprise Edition, a new platform that lets corporate security teams test whether their own artificial intelligence agents can actually perform the cybersecurity jobs they have been assigned, rather than relying on generic model benchmarks. The announcement landed on October 6, 2026, and was reported by SiliconANGLE, Bernama, and MSSP Alert, among others.
The move pushes Hack The Box (HTB), long known as a training ground for human penetration testers through Capture The Flag challenges and certification tracks, into a different business: certifying whether AI agents deployed inside real security operations teams can do the work companies are now paying them to do. It’s a timely pivot. As more firms hand off tasks like triage, log review, and even exploit development to autonomous agents, the gap between “the model scored well on a benchmark” and “the agent can reliably do this job” has become one of the more pressing questions in enterprise cybersecurity.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What AI Range Enterprise Edition Actually Does
AI Range itself isn’t brand new. HTB introduced the underlying AI Range concept on December 3, 2025, as what the company describes as a controlled cyber environment that simulates thousands of continuously updated offensive and defensive targets, built to test and benchmark the safety, limits, and capabilities of autonomous AI security agents, according to SiliconANGLE’s reporting. Hack The Box reported in December 2025 that AI Range offers new target and challenge releases every week, and by October 2026 AI labs, government customers, and hyperscalers were using it, according to IT Brief. The Enterprise Edition, launched on October 6, 2026, turns that testbed into something enterprise security teams can point directly at their own deployed agents.
The core idea, per coverage of the launch, is role-based appraisal: instead of grading an AI agent against an abstract leaderboard, AI Range Enterprise Edition assigns it a defined cybersecurity role and evaluates whether it can perform that role under realistic, pressure-tested conditions. At launch, AI Range Enterprise Edition could appraise AI agents against two defined cybersecurity job roles, penetration tester and security operations center (SOC) analyst, the first 2 roles Hack The Box named in October 2026, with additional roles planned for later rollout, according to reporting on the launch.
This matters because most organizations evaluating AI agents for cybersecurity work today still lean on generic model benchmarks or vendor-supplied demos, neither of which tells a CISO whether an agent connected to their actual environment, with their actual data and their actual threat landscape, will perform reliably six months from now. That disconnect is exactly what Hack The Box is positioning AI Range Enterprise Edition to close, similar in spirit to how enterprises already rely on dedicated tooling to lock down the credentials AI agents use to access internal systems.
Agentic Worker Competence and Agentic Operator Scoring
AI Range Enterprise Edition centers on Agentic Worker Competence, which measures whether an AI agent can carry out a defined cybersecurity job under realistic conditions inside the range. The offering also includes AI-augmented roles and a second scoring mechanism, Agentic Operator Competence Scoring, which does not benchmark AI agents. It evaluates the human side of the equation: whether a security professional can question AI output and intervene appropriately when directing, verifying, and correcting AI-assisted work, according to launch coverage.
That human-focused scoring is arguably the more interesting design choice, and it ships as part of the Enterprise Edition alongside the agent scoring. The upshot is that HTB evaluates the AI agent and the human operator within one offering, as separate measurements. Results come back as a role score alongside a pass-or-fail grade on individual environments, and customers can rerun the assessment whenever an agent, its underlying model, or its data changes, a reassessment feature MSSP Alert covered in October 2026, so organizations can track whether an agent’s performance improves, degrades, or holds steady over time.
That repeatability angle is a deliberate acknowledgment that today’s “good” AI agent might be tomorrow’s liability the moment its underlying model gets swapped, fine-tuned, or re-prompted. Security teams have already seen how fast this space moves; Google recently confirmed three separate incidents of AI agents acting outside their intended boundaries, a reminder of why continuous re-testing, not a one-time certification, is the actual value proposition here.
Where AI Range Enterprise Edition Fits in HTB’s Product Lineup
Hack The Box’s existing catalog is built almost entirely around human skill development: Capture The Flag challenges for competitive problem-solving, Pro Labs for advanced offensive-security practice across complex multi-machine environments, HTB Academy for structured course-based learning, Dedicated Labs for private organizational training environments, and the Certified Web Exploitation Expert (CWEE) credential for formal professional certification. AI Range Enterprise Edition is the first HTB product built to evaluate a machine rather than a person: it appraises AI agents against defined cybersecurity job roles. HTB’s separate AI-Augmented Roles offering is what assesses the human cybersecurity professionals working alongside those agents.
| HTB Product | What It Primarily Evaluates | Scoring Approach | Typical User |
|---|---|---|---|
| AI Range Enterprise Edition | AI security agents performing defined cybersecurity job roles | Role-based competence scores, pass/fail per environment, longitudinal tracking | Enterprise security teams, MSSPs, government agencies |
| Pro Labs | Human penetration testers | Completion of multi-machine attack chains | Individual practitioners, red teams |
| CTF Challenges | Individuals solving discrete security puzzles | Points and competitive leaderboards | Community members, students |
| HTB Academy | Structured security knowledge and skills | Course completion and module assessments | New entrants, upskilling teams |
| Dedicated Labs | Custom private training environments | Organization-defined exercises | Enterprise training and L&D teams |
| CWEE Certification | Professional-level exploitation expertise | Formal certification exam | Certified human pentesters |
HTB says its range ecosystem, the same one used to build and run AI Range, already serves more than 4 million human cybersecurity professionals and supports upward of 800 enterprise customers spanning Fortune 500 companies, government agencies, managed security service providers, and frontier AI labs, according to figures on the company’s own site reviewed as part of launch coverage. AI Market Watch reported that HTB’s available platform catalog exceeded 1,500 labs by October 2026. That installed base gives HTB a distribution advantage that a standalone AI-evaluation startup would need years to replicate.
Why the Timing Lines Up With a Broader AI Agent Reckoning
AI Range Enterprise Edition arrives in the middle of a run of stories about AI agents behaving unpredictably inside enterprise and government systems. Beyond Google’s disclosed agent escapes, the FTC has opened a multi-company probe into how leading AI labs deploy autonomous agents, and lawmakers have floated accountability legislation partly in response to incidents tied to AI agent failures. Those stories share a common thread with HTB’s pitch: organizations are deploying AI agents faster than they’re building the testing infrastructure to verify those agents behave as intended, let alone perform their assigned jobs competently under pressure.
That regulatory backdrop is not incidental to how HTB is framing the product. Lawmakers pushing the AI Agent Accountability Act have specifically cited the lack of evidence that deployed agents can perform their intended functions safely, and the FTC’s inquiry into OpenAI and Anthropic’s agent deployments turns on similar questions about operational reliability versus marketed capability. A product that produces a documented, repeatable competence score for a deployed agent gives enterprise buyers something concrete to point to if a regulator, insurer, or board member asks whether their AI workforce has actually been tested.
What Hack The Box Leadership Is Saying
Hack The Box founder and CEO Haris Pylarinos framed the launch around the idea that AI doesn’t eliminate the need for human expertise in security operations, it relocates where that expertise is most valuable. In comments tied to the Enterprise Edition launch, Pylarinos said: “AI does not remove the need for cyber expertise. It changes where that expertise matters most. The strongest security teams will not be the ones that simply deploy the most AI. They will be the ones with people who have the judgment to direct it, verify its work, and step in when needed. And with agents that are held to the same standards as the people they work beside. Cyber readiness now means building confidence in both.” (Hack The Box blog)
Pylarinos struck a similar note in comments reported by Bernama around the Enterprise Edition rollout: “As AI takes on more cybersecurity tasks, leaders need confidence in both sides of the workforce. They need to know that agents can do the jobs they are being given, but they also need people with the expertise and judgement to direct that work, verify it and step in when needed.” (Bernama)
On the original AI Range launch that the Enterprise Edition builds on, Pylarinos described the company’s broader ambition this way: “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.” (Hack The Box blog)
Hack The Box Chief Product Officer Gerasimos Marketos put the product’s philosophy in blunter, more operational terms: “Hack The Box is where AI agents and humans learn to operate under real pressure together.” (Hack The Box blog)
The AI Security Skills Gap Behind the Product Bet
HTB’s commercial argument rests on a workforce reality that’s been building for several years: security teams are short-staffed, and AI agents are being pitched as a way to stretch thin teams further. The risk HTB is explicitly selling against is that a company deploys an agent after a vendor demo or a generic benchmark run, never checks whether it can continuously perform its actual assigned role as models get updated, threat patterns shift, and attack surfaces grow, and only discovers the gap after an incident.
That’s a familiar pattern in enterprise security more broadly. Teams routinely run static analysis and dependency scanning as a baseline check on code before it ships, the same logic driving adoption of tools like Semgrep for automated security scanning, precisely because waiting to discover a vulnerability in production is far more expensive than catching it earlier in a controlled environment. AI Range Enterprise Edition is HTB’s attempt to apply that same “test before you trust it in production” discipline to AI agents doing security work rather than to code.
Competitive Landscape: Who Else Is Testing AI Security Agents
Hack The Box isn’t the only company racing to put guardrails and scorecards around AI agents in security contexts, though the field remains fragmented across several distinct approaches rather than consolidated around one standard. Vendors focused on AI red-teaming and model evaluation probe a model’s or agent’s safety boundaries directly, looking for prompt-injection weaknesses or unsafe tool-use behavior. A separate category, AI agent governance and permissioning platforms, controls what an agent is allowed to touch rather than measuring whether it’s good at its job. Traditional breach-and-attack simulation vendors approach the problem from the defensive-infrastructure side, validating whether existing controls hold up against simulated attacks, which is a related but distinct question from whether a specific AI agent can perform a specific role competently.
| Category | What It Tests | How It Differs From AI Range Enterprise Edition |
|---|---|---|
| AI red-teaming / model evaluation | Model safety boundaries, prompt-injection resistance, unsafe tool use | Tests the model in isolation, not role-based job performance inside a live range |
| AI agent governance platforms | What systems and data an agent can access | Permissions layer, not a competence or skills measurement |
| Breach-and-attack simulation | Whether existing defensive controls withstand simulated attacks | Measures infrastructure readiness, not a specific agent’s role competence |
| Traditional cyber ranges (human-focused) | Human analyst and pentester skills | HTB’s own legacy category; AI Range extends the same range model to agents |
| AI Range Enterprise Edition (HTB) | Combined agent-plus-operator competence on defined cybersecurity roles | Only entrant combining range realism, role-based scoring, and repeat testing over time |
Security teams evaluating AI agent risk more broadly have also started comparing dedicated AI agent security vendors against each other on funding and market traction, as seen in the ongoing scrutiny of players like Zenity, HiddenLayer, and Straiker in the AI agent security funding race. Hardware vendors are getting pulled into the same conversation; NVIDIA has bet on dedicated BlueField-4 silicon for AI agent safety enforcement, approaching the same underlying problem, trustworthy AI agents in security-sensitive environments, from the infrastructure layer rather than the training-range layer HTB occupies.
From CTFs to Agent Appraisal: A Short History of Hack The Box
Hack The Box built its reputation over roughly a decade as a Capture The Flag and penetration-testing practice platform aimed squarely at human hackers, growing into one of the more recognizable brands in offensive-security training. Its expansion into enterprise-facing products, Dedicated Labs, professional certifications like CWEE, and corporate training contracts, tracked the broader industry shift toward formalized, continuously updated security skills programs rather than one-off training events.
HTB launched Threat Range on September 30, 2025, adding team-based incident simulations for enterprise security operations. The original AI Range, introduced in December 2025, marked the company’s first explicit move to apply its range methodology to machines instead of people. AI Range Enterprise Edition, roughly ten months later, represents the commercialization step: taking a proof-of-concept testing environment and turning it into a product enterprise security teams can buy, deploy against their own agents, and use to generate ongoing compliance and operational evidence. That’s a familiar arc in enterprise software, research tool first, enterprise product second, and it suggests HTB sees sustained demand rather than a one-off news cycle around AI agent testing.
Market Impact: What This Means for Security Budgets
For enterprise security buyers, AI Range Enterprise Edition adds a new line item to an already crowded AI-governance budget category: agent evaluation and continuous re-certification. That’s a cost center that barely existed eighteen months ago. HTB did not publish a numeric price for the Enterprise Edition in its launch materials, with pricing details requiring direct inquiry through the company’s sales channel rather than its public pricing page, according to a review of HTB’s own site alongside launch coverage.
The absence of public enterprise pricing is fairly typical for HTB’s higher-tier offerings; its publicly listed plans on the Hack The Box pricing page cover individual and team subscriptions, while enterprise and Dedicated Labs contracts have historically been negotiated directly. That pattern is likely to hold for AI Range Enterprise Edition, particularly given that the roles and scoring methodology are still expanding beyond the initial penetration-tester and SOC-analyst coverage.
Cyber insurers are a less obvious but potentially significant customer segment here. Insurance underwriters evaluating cyber risk have increasingly asked about the controls a company has in place around automated systems, a trend visible in how firms like Coalition, Chubb, and At-Bay price cyber coverage. A documented, repeatable AI agent competence score is exactly the kind of artifact an underwriter could eventually ask to see before pricing a policy that covers AI-related incidents.
Open Questions and Risks in Scoring AI Agents
A few open questions hang over the launch that the available coverage doesn’t yet resolve. It’s unclear how AI Range Enterprise Edition’s scoring methodology will be independently validated, whether a third party will audit the pass/fail criteria, or whether the two launch roles (penetration tester and SOC analyst) will expand quickly enough to cover the much longer list of jobs companies are already assigning to agents, from incident response to cloud configuration review to vulnerability triage.
There’s also a harder structural question: a competence score measured today against today’s threat environment says little about an agent’s behavior against a threat pattern that doesn’t exist yet. HTB’s own materials acknowledge this by building in repeat testing as models, agent software, and threat conditions change, but that design choice also means AI Range Enterprise Edition is a subscription commitment to ongoing re-testing, not a one-time stamp of approval, which is a meaningfully different sell than a traditional certification exam.
Predictions: Where AI Agent Security Testing Goes Next
- More roles, faster than the current two. Expect HTB to expand past penetration tester and SOC analyst into incident response, cloud security, and GRC-adjacent roles within the next few quarters as enterprise customers push for broader coverage.
- Agent competence scores become procurement requirements. Large enterprises and MSSPs are likely to start asking AI agent vendors for third-party competence evidence before contracts are signed, mirroring how SOC 2 reports became table stakes for SaaS procurement.
- Rival cyber range and training vendors respond within two quarters. HTB’s installed base of 800-plus enterprise customers gives it a head start, but competitors in adjacent categories, from breach-and-attack simulation to AI red-teaming, have clear incentive to bolt on comparable role-based scoring.
- Cyber insurers start referencing agent testing in underwriting. As AI agent incidents accumulate, expect insurers to begin asking policyholders whether deployed agents have been competence-tested, much as they already ask about MFA and endpoint detection coverage.
- Regulatory scrutiny adds pressure rather than replacing it. Legislative efforts like the AI Agent Accountability Act and ongoing FTC inquiries are unlikely to mandate a specific testing standard soon, but they raise the cost of having no evidence at all, which strengthens the commercial case for products like AI Range Enterprise Edition regardless of how the legislation ultimately lands.
What Security Leaders Should Do Next
For CISOs and security engineering leads already running AI agents in production, or planning to within the next year, the practical takeaway from this launch isn’t necessarily “buy AI Range Enterprise Edition.” It’s that the industry is converging on the idea that agent deployment needs its own testing discipline, distinct from both model benchmarking and traditional penetration testing. Teams that have invested in securing the credentials and access scopes their AI agents operate under have already done half the groundwork; competence testing against defined roles is the logical next layer on top of that access-control foundation.
Whether that testing comes from Hack The Box, a competing range vendor, or an internally built evaluation harness, the direction of travel looks set: generic benchmark scores and vendor marketing claims are increasingly treated as insufficient evidence on their own for high-stakes AI agent deployments in security operations.
Frequently Asked Questions
What is Hack The Box AI Range Enterprise Edition?
It’s a platform launched by Hack The Box on October 6, 2026, that lets enterprise security teams test and score whether their own AI security agents can perform defined cybersecurity roles under realistic, controlled conditions.
How is it different from the original AI Range?
HTB introduced AI Range in December 2025 as a controlled environment for benchmarking AI agent safety and capability. Enterprise Edition builds on that foundation by adding role-based appraisal, Agentic Worker Competence and Agentic Operator Competence Scoring, and tools for enterprises to test their own deployed agents directly.
What AI-augmented roles are available at launch?
Two roles launched initially: penetration tester and security operations center (SOC) analyst. HTB has indicated additional roles are planned for future rollout, according to launch coverage.
How much does AI Range Enterprise Edition cost?
HTB has not published a specific price for the Enterprise Edition. Enterprise pricing typically requires direct inquiry through HTB’s sales channel rather than being listed on its public pricing page.
Who is the target customer for this product?
HTB describes its broader enterprise base as including Fortune 500 companies, government agencies, managed security service providers, and frontier AI labs, and positions AI Range Enterprise Edition toward security teams in those same segments that are actively deploying AI agents into operational workflows.
Does AI Range Enterprise Edition replace human security analysts?
No. HTB’s own framing, echoed by CEO Haris Pylarinos, explicitly positions the product as scoring both the AI agent’s competence and the human operator’s judgment in directing and verifying that agent’s work, not as a replacement for human expertise.
How does this relate to other Hack The Box products like Pro Labs or CWEE?
Pro Labs, CTF content, HTB Academy, Dedicated Labs, and the CWEE certification are all built to develop and validate human skills. AI Range Enterprise Edition is HTB’s first product built to evaluate AI agents and AI-assisted human-agent teams specifically.
Are there competing products that do the same thing?
Not directly in the same combined format. AI red-teaming vendors, AI agent governance platforms, and breach-and-attack simulation tools address related but distinct problems. No other named competitor currently combines range-based realism, role-specific scoring, and repeatable longitudinal testing in one enterprise product, based on available launch coverage.
