Adversarial exposure validation
Adversarial exposure validation is Gartner’s name for tools that give continuous, automated evidence that an attack could work against you. It folds breach and attack simulation and automated pentesting into one category.
What it proves
That a given exposure is or is not usable by an attacker in your environment, with your controls in place. A critical CVE behind a working EDR and a segmented network may be harmless. A medium-rated misconfiguration on a path to a domain admin may not be.
Gartner published its first Market Guide for Adversarial Exposure Validation on 11 March 2025 and an update on 24 March 2026. Gartner Peer Insights describes the category as replacing breach and attack simulation and automated penetration testing and red teaming.
What it does not prove
AEV is a label, not a feature set. Two platforms sold as AEV can test very different things. One may run thousands of control scenarios and a light lateral-movement module. Another may chain real exploits through your network and barely touch email or web controls.
It also does not discover assets on its own in most products. It validates exposures that a scanner, an ASM tool or a CTEM programme has already found.
How it differs from its neighbours
Breach and attack simulation is the control-testing half. Automated pentesting is the exploit-and-pivot half. AEV is the claim to do both. Exposure management platforms sit one step earlier: they collect and rank exposures, and some model attack paths rather than test them.
Who buys it
Security teams running a CTEM programme who need the "validation" step to be more than a spreadsheet. Usually a mix of SOC, vulnerability management and red team people, reporting to a CISO who wants fewer, better-argued fixes.
Because the label is broad, the pilot matters more here than for any other type. Read the vendor's own description of each half before you book a demo.
Vendors that fit
Listed alphabetically: documented first, then partial claims. The note under each name is a summary of what the vendor's own pages claim, as read on 3 November 2026.
Documented (6)
- AttackIQ (documented)AttackIQ's AWS listing and Flex FAQ describe breach and attack simulation; DISA's selection names it an AEV platform; the Exposure Management Module adds attack-path modeling and risk scoring; Enterprise is described as a 'prescriptive service' run with AttackIQ experts (managed/co-managed); 'offensive testing' of full attack paths is marketed on the homepage, but no stand-alone automated pentest product was found.
- Cymulate (documented)Cymulate's AWS listing describes an exposure validation platform 'that includes breach and attack simulation and automated red teaming'; the product whitepaper documents Attack Path Discovery (lateral movement via Pass-the-Hash, Kerberoasting, password spraying) and CTEM prioritization with a validated exposure score; no dedicated 'automated penetration testing' product or managed validation service was found on the pages read.
- Pentera (documented)Pentera's own pages describe AI-driven internal, external and cloud pentesting (Core, Surface, Cloud), attack path root-cause analysis and Resolve remediation workflows supporting CTEM, and it was named a 2026 Gartner Peer Insights Customers' Choice for AEV; security control assessment (EDR, SIEM, firewalls) happens inside attack runs rather than through a separate BAS scenario library; human-led red-team services come via the 2025 EVA acquisition rather than a managed validation subscription.
- Picus Security (documented)Picus's July 2026 launch release names three products (Breach and Attack Simulation, Autonomous Penetration Testing, Exposure Validation) under one AEV platform; the APV datasheet documents attack path mapping to domain admin; EXV ingests scanner findings (Tenable, Qualys, Wiz) and Attack Surface Validation adds CAASM; Picus sells 100% through channel partners including MSSPs, but no Picus-run managed service was found.
- SafeBreach (documented)SafeBreach's own pages call Validate a BAS tool and Propagate 'an enterprise-grade automated penetration testing tool' for attack path validation, and brand the combination as AEV; the CTEM Platform adds Helm agents for scoping, discovery and prioritization but relies on third-party VM and EASM data, with its own Exposure Hub announced as 'upcoming' in April 2026; no SafeBreach-run managed service was found (a Ness professional-services offer on AWS Marketplace is third party).
- SCYTHE (documented)SCYTHE's own pages position it as Adversarial Exposure Validation that evolved from BAS (5.0 release) and list Managed AEV and Managed Purple Teaming services on the pricing page; the AI campaign builder generates attack sequences, but no attack-path graph or exposure-management module is claimed.
Partial (4)
- Horizon3.ai NodeZero (partial)Horizon3 positions NodeZero as continuous autonomous penetration testing that chains and exploits weaknesses with proof; detection-side features (Tripwires) and threat-driven Rapid Response tests extend it toward exposure validation, but it does not claim BAS-style control-by-control simulation or a managed service.
- Mandiant Security Validation (partial)Google Cloud describes the product as testing security controls with breach and attack emulations; continuous drift testing (AEDA) supports part of the exposure-validation use case, but no attack-path or automated pentest capability is claimed on the product page.
- Validato (partial)Validato describes itself as a BAS / continuous security control validation platform and its FAQ describes it as using Adversarial Exposure Validation; it says it uses simulation of ATT&CK behaviours rather than payload emulation, and no attack-path, pentest or managed service is claimed.
- XM Cyber (partial)XM Cyber's pages position it as Continuous Exposure Management with Attack Graph Analysis; its 'validation' is done by modelling exploitability and reachability in a digital twin ('safely simulate attack paths without impacting production'), not by executing attack scenarios against live controls, so BAS and automated pentesting are not claimed.