Hack The Box is launching AI Range Enterprise Edition to let security teams assess their own AI agents on the roles assigned to them.[1] For businesses, the stakes are concrete: having evidence before delegating a task, then checking that this evidence remains valid as tools evolve.
An enterprise edition to assess defined missions
AI Range has existed since December 2025. The novelty announced on 6 October 2026 is an edition aimed at corporate security teams, with role-based assessment. Hack The Box presents this approach as a way of regularly measuring an agent’s ability to do the work expected of it.[1][3]
A cyber range is a controlled training environment. It allows tools or professionals to be confronted with representative situations without a real incident being the first test. Passing these exercises provides evidence for assessment, the scope of which depends on the scenario and the conditions chosen.
An overall score is not enough to assign responsibility
An agent may recognise suspicious activity without knowing how to build a complete investigation. For a company, defining its mission therefore means specifying the information it can access, the decisions it is allowed to make and the situations that require human validation.
Take an illustrative case: an assistant triages login alerts. The assessment should check the incidents identified, but also the legitimate alerts dismissed and the cases left unanswered. A quick result loses its value if an analyst then has to redo the whole investigation.
Tests must keep pace with changes to the agent
Hack The Box recommends reassessing agents whenever the models, the components that drive their actions, the environments or the threats change.[2] Otherwise, a one-off validation risks describing a configuration that is no longer the one used day to day.
In practice, the team should keep a record of the version tested, the tools available, the permissions and the results. This makes it possible to compare two assessments and determine whether a change genuinely improves the service.
Four criteria for examining the work delegated
This grid is an editorial proposal for preparing an assessment, not Hack The Box’s scoring scheme.
| Criterion | Result to observe | Control question |
|---|---|---|
| Relevance | Incidents detected and false alarms. | Which important cases are missed? |
| Justification | Sources and evidence supporting the analysis. | Can an analyst verify the conclusion? |
| Permissions | Actions carried out within the intended scope. | When must the agent ask for validation? |
| Cooperation | Handing uncertain cases over to a human. | Does the escalation contain the useful information? |
The analyst’s judgement is also put to the test
The announcement also introduces Agentic Operator Competence Scoring, a proprietary system separate from agent assessment. It aims to measure professionals’ ability to understand a recommendation, spot an error and decide when to step in. The AI-augmented SOC Analyst role uses AI-assisted alert analysis and investigation scenarios.[1]
This distinction avoids reducing oversight to systematic approval. A professional must be able to challenge a conclusion, but also to let an authorised operation continue when the available evidence justifies it.
For SMEs, start with a measurable task
An SME could assess an agent on preparing alert files before considering letting it act on its systems. The pilot should compare the quality of the files and the total time spent by the team, including corrections. Pricing, integrations and access conditions for the offering remain to be confirmed with the vendor.
The press release describes a method and announced capabilities. It does not demonstrate a reduction in incidents for every customer. A laboratory result must be complemented by validation in the company’s own context, with limited rights and defined stop criteria.[1]
Professionals become responsible for the quality of delegation
This development gives concrete form to Cybersecurity. Professionals must be able to design a relevant test, examine the traces produced and recognise a poorly supported conclusion. Understanding threats remains necessary to judge the agent’s work.
Their role also involves setting the limits of delegation. They must distinguish a recommendation from an authorised action and organise a handover to humans when information is missing or the risk goes beyond the intended scope.
The value of this skill is then measured in the quality of the decisions made with the tool. Being able to explain why a result is acceptable, or why it must be rejected, becomes part of the defence work.
Trust that must be renewed with use
Our reading of this announcement is that agent assessment could become a regular operational activity. The challenge will be to link each result to a specific mission and a deployment decision. The useful question remains simple: what evidence currently justifies the autonomy granted to this agent?
ALSO READ
These four CCCLX analyses explore further the issues of permissions, protection and system verification.
GitHub Copilot bulks up for business: coding agents come under IT control
VAST Data wants to let AI use sensitive data in a protected environment
Project Glasswing: Anthropic goes on the offensive against software vulnerabilities
ShinyHunters bypasses WAF rules and revives attacks on PeopleSoft
Sources
- 01
Hack The Box (6 October 2026). “Hack The Box Launches AI Range Enterprise Edition to Set a New Standard for AI Agent Competence in Cybersecurity”. Presentation of the enterprise edition and role-based agent assessment.
- 02
Hack The Box. “Cyber Workforce Development for Human & Agentic Teams”. Presentation of the offerings and continuous agent reassessment.
- 03
Hack The Box (3 December 2025). “Hack The Box launches the world’s first AI cyber range to benchmark AI agents and accelerate human-AI teaming across offensive and defensive cyber operations”. Initial announcement of AI Range.
Checked on 6 October 2026: the Hack The Box sources were consulted via their indexed content, as direct access was unavailable. The four CCCLX links were found; the VAST and PeopleSoft pages were checked in their indexed content after direct access failed. The criteria in the table are editorial and do not reproduce the proprietary scoring scheme.
Salesforce wants to acquire Loops to help Tableau explain KPI changes
Salesforce has announced a definitive agreement to acquire Loops, an AI-based analytics platform, with the intention of integrating its technology into Tableau after closing.[1] The aim is to monitor metrics and bring out the factors associated with their changes. For...
Cisco and NVIDIA prepare 400 Gb/s-per-lane optical connections for AI infrastructure
Cisco and NVIDIA have announced a project for a pluggable optical form factor, USFP, capable of supporting 400 Gb/s per lane. Cisco is also unveiling the Silicon One G302 switching chip, rated at 51.2 Tb/s.[1] These announcements address a key challenge for AI...
A new year begins at CCCLX!
The Paris-Villejuif campus welcomes CCCLX students for a new year of learning, projects and experiences at the heart of 360° IT.
VMware HyperArmor strengthens hypervisor protection against cyberattacks
Broadcom has unveiled VMware HyperArmor, a solution designed to protect the ESX hypervisor of VMware Cloud Foundation, with announced support for ESX 8.x and later versions.[1] The stakes go beyond the security of a single application: the hypervisor is a shared...
SolarWinds Avior: IT teams set the goals and agents coordinate the actions
SolarWinds has launched Avior, an agentic AI system designed to pursue operational goals in hybrid IT environments. Available since 14 October as part of a guided onboarding programme, it combines specialised agents with control mechanisms.[1] For technical teams, the...
GitHub steps up vulnerability detection and remediation with AI Scan and Agentic Autofix
GitHub has announced the general availability of AI Scan for pull requests and Agentic Autofix for code scanning alerts.[1][2] These features bring AI assistance closer to the work of securing code. They do, however, operate on different scopes: proposed...
Cloudflare makes web security programmable with Workers
Cloudflare is bringing its secure web gateway closer to its Workers development platform. With Forward, available in public beta, businesses can add code to the processing of certain authorised HTTP requests.[2] For IT teams, this development opens up a...
Palo Alto Networks strengthens AI agent security around Claude Enterprise
PPalo Alto Networks has unveiled several updates to Prisma AIRS: protection of interactions with Claude Enterprise, analysis of the components used by agents and control of their exchanges with tools.[1] The stakes are concrete for businesses: preventing an assistant...
ServiceNow prepares AI Passport to control AI agents’ access rights
ServiceNow has announced AI Passport, a cryptographically signed authorisation credential designed to govern AI agents and MCP servers while workflows are running. It is scheduled to be available in the fourth quarter of 2026.[1] The image of a digital badge helps to...
SAP wants to acquire TechWolf to link tasks, skills and the transformation of jobs
SAP has announced an agreement to acquire TechWolf, an AI platform that links tasks, skills and changes in the labour market. Completion is expected in the fourth quarter of 2026, subject in particular to regulatory approvals.[1] The deal raises a practical...
