METHODOLOGY & INTERNATIONAL REFERENCES

AI-Resilience Web Assessment Methodology

AVALUE AI.RESWA is an independently developed, evidence-based methodology for assessing observable web resilience against AI-assisted and autonomous attack behavior through bounded, non-destructive external testing.

01

Recon Resistance

Measures public attack-surface intelligence, artifact exposure, metadata, error disclosure, technology leakage and route behavior.

02

Exploit Resistance

Observes safe input, HTTP, parser and browser-policy boundaries without destructive exploitation.

03

Automation Resistance

Evaluates bounded repeated requests, bot identity, rate-control signals and observable automation friction.

04

Adaptive Resistance

Checks response stability against harmless case, encoding, path and query mutations.

05

Chaining Resistance

Assesses observable prerequisites that could help an autonomous attacker combine minor exposures into a larger attack path.

06

Detection & Recovery

Observes public defensive signals while clearly separating externally verifiable evidence from internal operational readiness.

Measurement, correlation and evidence

The 0–100 AI.RESWA score is an independent six-domain measurement. Evidence Coverage is a separate confidence measure. AVALUE WebSec independently assesses foundational web security and produces grade A, B, C, D or F. Neither result is used to alter the other.

0–29 · Critical Exposure30–44 · High Exposure45–59 · Elevated60–74 · Moderate75–89 · Strong90–100 · Advanced

The public LEVEL and Autonomous AI Attack Risk are integrated outputs produced by correlating the AI.RESWA measurement band with the independent AVALUE WebSec grade. Public LEVEL uses EXCELLENT, STRONG, MODERATE, GUARDED, POOR, WEAK and CRITICAL. The paired risk classes are Very Low, Low, Moderate, Elevated, High, Very High and Critical. The correlation model is an AVALUE methodology and is not an OWASP or NIST rating system.

Why can the score and LEVEL appear different?

The AI.RESWA score is an independent measurement of six AI-resilience domains, while the public LEVEL is an integrated assessment that also considers foundational web security through AVALUE WebSec. A site can therefore have a relatively good AI.RESWA score but receive a lower LEVEL when WebSec identifies material foundational weaknesses. This reflects the principle that fundamental security weaknesses can increase the opportunity for successful autonomous AI attacks.

WebSec does not alter the AI.RESWA score. The two independent results are correlated only to determine the public LEVEL and Autonomous AI Attack Risk.

INTERNATIONAL REFERENCE MAPPING

Methodological grounding

OWASP Web Security Testing Guide (WSTG) v4.2

Primary reference for established web application security testing practices and structured testing scenarios.

Official OWASP reference ↗

OWASP Automated Threats to Web Applications

Reference for technology-agnostic automated threat events such as fingerprinting, footprinting, scraping and vulnerability scanning.

Official OWASP reference ↗

NIST Cybersecurity Framework (CSF) 2.0

Reference for understanding, assessing, prioritizing and communicating cybersecurity risk and resilience outcomes.

Official NIST reference ↗

NIST AI 100-2e2025

Conceptual reference for adversarial AI terminology, attacker goals, capabilities and attack lifecycle. It addresses attacks on AI systems and is not itself a web-resilience testing standard.

Official NIST reference ↗
Independent Methodology Notice

AVALUE AI.RESWA is independently developed. References to OWASP and NIST indicate methodological alignment or conceptual grounding and do not imply endorsement, certification or affiliation by those organizations.