Short answers to the questions teams ask most before their first AI Network Pentesting assessment.

## How Is This Different from AI Pentesting?

Both run on [Cascade](<https://docs.escape.tech/documentation/ai-pentesting/multi-agent-pentest/index.md>). AI Network Pentesting targets IPs, ranges, and domains, and tests the services listening on them: exposure, configuration, TLS, DNS, and credentials. [AI Pentesting](<https://docs.escape.tech/documentation/ai-pentesting/index.md>) targets one web app or API, signs in as your users, and tests access control and business logic. Run both: the network assessment shows what's exposed, the AI Pentesting assessment tests the apps behind it.

## How Is This Different from ASM?

[ASM](<https://docs.escape.tech/documentation/asm/index.md>) keeps an inventory of your exposed assets and flags issues on them. AI Network Pentesting takes a fixed scope and attacks it, then reproduces each finding with proof.

## Does It Test Internal Networks?

Yes. Deploy a [private location](<https://docs.escape.tech/documentation/platform/private-location/index.md>) inside your network and select it in **Fine-Tune (Optional) \> Location**. The assessment then reaches your private ranges and internal domains the way an attacker who got inside would. Active Directory isn't covered yet.

## Does It Find My Subdomains for Me?

Not yet. It tests the domains, addresses, and ranges you enter. List every subdomain you want tested.

## Does It Test Firewalls, Routers, or VPN Gateways?

It tests what they expose on the IP addresses in scope: open ports, administration interfaces, TLS settings, and credentials on services that authenticate. It doesn't review device configuration from the inside.

## What If My Services Sit Behind a Reverse Proxy?

Escape tests what an external attacker reaches, which is the proxy and whatever it forwards. Services that the proxy doesn't expose aren't reachable, so they aren't tested.

## Will It Break Anything?

Safe vectors only observe. Intrusive vectors can lock accounts, fill logs, or slow a service, and they're labeled as such in the form. Every agent follows non-destructive rules whatever you select. See [Attack Vectors](<https://docs.escape.tech/documentation/network-pentesting/attack-vectors/index.md>) and [Guardrails](<https://docs.escape.tech/documentation/network-pentesting/guardrails/index.md>).

## How Long Does an Assessment Take?

You set the maximum: 6 to 24 hours, 10 hours by default. See [Max Duration](<https://docs.escape.tech/documentation/network-pentesting/guardrails/#max-duration>).

## How Do I Get Access?

AI Network Pentesting is a separate product in beta. Contact your Escape representative to enable it.