AI Security and Resilience: Testing an AI system against adversarial input

Auditors ask about ai security and resilience early, because the answer tells them how the rest of the program is likely to look. The requirement is NIST AI RMF MEASURE 2.7.

What the rule says

NIST AI RMF MEASURE 2.7 sets the expectation for ai security and resilience. The framework describes the outcome rather than prescribing a technology, so the question is whether your implementation achieves it and whether you can show that it does.

What this looks like in practice

In an assessment, NIST AI RMF MEASURE 2.7 is not one question. It resolves into several, and each one is really asking for a different artifact:

• Assess AI systems for security vulnerabilities unique to AI and machine learning

Ownership usually sits with the IT or security team, though the evidence is often produced by someone else, which is where the trail breaks. A control that works but has no owner tends to stop working the month the person who quietly maintained it changes roles.

Who this applies to

NIST AI RMF is voluntary rather than binding, and that changes how it should be used. Adopting it does not create a legal obligation, and declining it is not a violation. It earns its place when a customer contract names it, when it gives structure to a program HIPAA describes only in outcomes, or when a recognized practice is worth having on record. Treat it as a way of organizing work you already owe, not as a second regime.

Building it, in order

The sequence below is the order that produces evidence as a by-product rather than as a separate documentation exercise:

1. Assess each AI system for security weaknesses, including risks such as adversarial manipulation, data poisoning, and model or prompt exploitation.
2. Document identified vulnerabilities and their potential impact.
3. Define and track remediation for significant findings.
4. For vendor-supplied AI systems, obtain the vendor's security testing and assessment documentation and spot-check outputs rather than performing in-house adversarial testing.
5. Upload the AI security vulnerability assessment.

What else this satisfies

The same evidence answers more than one framework. The questions behind NIST AI RMF MEASURE 2.7 also map to NIST 800-53 RA-5, SA-11; HITRUST 10.m; ISO 42001 A.6.2.4. That matters for scoping: if you are working toward SOC 2 or an ISO certification alongside HIPAA, this control is one piece of work and several answers, provided the artifact is written once and referenced rather than rewritten per framework.

Evidence an assessor asks for

For this requirement the artifact types that satisfy it are report. The distinction matters more than it looks: a policy states what you intend to do, a procedure states how, and a record proves it happened on a date. Auditors ask for all three, and a practice that has written the first two often has nothing for the third.

Date every artifact and keep the superseded versions. No federal rule sets a retention period for a voluntary framework artifact, so this is our recommendation rather than a requirement: keep six years, because that is what HIPAA requires of the documentation this work usually overlaps with (45 CFR 164.316(b)(2)(i) for security documentation, 164.530(j)(2) for privacy), and holding two sets of dates is how the earlier one goes missing.

Where this usually goes wrong

The usual gap is a control that exists in practice and nowhere in writing. Someone does the work, reliably, and it has never been written down, so the organization cannot demonstrate it and cannot notice when it stops. The fix is not more control, it is a dated record and a named owner.

What to do about it

• Produce the artifact, not the intention. For NIST AI RMF MEASURE 2.7 that means a dated report with the period it covers and who reviewed it. If the only evidence is that someone knows how this works, you have a practice and not a record.
• Put a name on it. Whoever administers the systems should be recorded as the owner. A control with no owner stops working the month the person who quietly maintained it changes roles, and nobody notices.
• Set a review cadence and keep the evidence of each review. An annual date with a dated output is enough; what fails an assessment is a control implemented once and never re-checked against a changed environment.
• Do not treat this as a legal obligation. It is a voluntary framework. Adopt it where a customer contract names it or where it gives structure to something HIPAA describes only as an outcome, and say in your own documentation which of those reasons applies.

The recurring lesson with ai security and resilience is that documentation is not the paperwork after the control. It is the part an assessor can read, and therefore the part that has to exist.

References

• MEASURE 2.7, in NIST AI Risk Management Framework 1.0 (NIST AI 100-1). This is a voluntary framework, not a legal requirement.
• Crosswalk references in this post are our own mapping and are not verified against the published text of those standards.
• This post describes what a requirement asks for and what evidence documents it. It is not legal advice, and your counsel owns the conclusion. State law may impose stricter obligations (45 CFR 160.203).