Version: 2026-q2 Tier: v1 Preview (46 questions) Purpose: Methodology sample — NOT a real evaluation Valid through: June 30, 2026 Next release: July 1, 2026 (Q3 2026) License: CC-BY-4.0 Publisher: Memoriant, Inc.

What This Is

A 46-question preview of the Memoriant Industrial Benchmark methodology. This is NOT a full evaluation — it's a small sample intended to show you how Memoriant structures compliance AI testing.

v1 is the smallest tier in the Memoriant benchmark family. It exists to give researchers, AI builders, and compliance teams a quick taste of the benchmark methodology before they commit to running the full evaluation.

Do not mistake v1 for a credible AI evaluation tool. 46 questions cannot validate a model for production use in any regulated industry. This preview is educational, not evaluative.

⚠️ AI Safety Disclaimer — Always Review Output

AI systems make mistakes. Always review AI-generated output before using it for any purpose.

This benchmark is used to evaluate AI systems that generate compliance guidance. Even AI systems that score highly on this preview benchmark can produce:

Factually incorrect information — even with high benchmark scores
Hallucinated citations — references to regulations, controls, or documents that do not exist
Outdated guidance — AI knowledge reflects training cutoff, not current regulations
Confident errors — AI often states wrong information with the same confidence as correct information
Plausible-sounding fabrications — responses that read like expert advice but are invented

Before using any AI output for:

Compliance documentation (SSPs, POA&Ms, audit responses)
Regulatory submissions to DoD, NIST, or other agencies
Internal policy or procedure creation
Assessment preparation or C3PAO engagements
Legal or contractual decisions

You must:

Have a qualified human review every output
Verify citations independently against authoritative sources
Cross-check against NIST publications, DoD guidance, Federal Register
Document the review process for audit purposes
Never submit AI output directly — AI drafts are starting points, not finished products

Memoriant's position: AI is a force multiplier for compliance professionals, not a replacement. The human stays accountable. The AI accelerates the work.

What You Should NOT Do With v1

Do not draw conclusions about model quality from this sample. 46 questions cannot validate a model for production compliance use. Statistical significance requires the full benchmark.
Do not publish benchmark scores using only v1. Scoring a model on v1 alone produces noise, not signal. It is not a credible evaluation for marketing or research claims.
Do not use this to decide whether to deploy compliance AI. You will miss critical failure modes that only show up in larger evaluation sets.
Do not cite v1 as "the CMMC benchmark" — it is explicitly labeled a preview for a reason.

What v1 Is For

Understanding the Memoriant benchmark methodology
Quick sanity check during early model development
Educational use in AI research contexts
Demonstrating what "compliance AI evaluation" looks like
Testing benchmark harness integration before committing to larger evaluations

⚠️ Version Expiration

Valid through: June 30, 2026 Next release: July 1, 2026 (Q3 2026)

This benchmark is dated. CMMC regulations, DFARS clauses, and NIST publications update continuously. Testing AI systems against a frozen benchmark produces false confidence — models that pass today's questions may fail tomorrow's.

If you are using this benchmark after the expiration date, your evaluation is incomplete. The Memoriant benchmark family is refreshed quarterly to incorporate:

New DFARS clauses and amendments
NIST SP 800-171/172 revisions and errata
CMMC Program Office guidance updates
New attack patterns and compliance failure modes
Emerging regulatory questions from the field

To get the latest version: Follow memoriant on HuggingFace, or subscribe to release notifications at memoriant.ai.

For Real Evaluation

If you need to actually evaluate a compliance AI system, use one of these instead:

Benchmark	Purpose	Questions
v1 Preview (this)	Methodology sample — not for validation	46
v2 Spot Check	Triage tool — eliminates obvious failures	454
v3 Comprehensive	The authoritative standard — use this for real evaluation	1,273

If your AI is going into a defense compliance workflow, v3 is the only credible evaluation.

Usage

from datasets import load_dataset

# Requires login to HuggingFace and auto-approved access
dataset = load_dataset("memoriant/cmmc-benchmark-v1-preview-2026-q2")
test = dataset["test"]  # 46 questions

# Example record structure
{
  "id": "T1-Q01",
  "tier": 1,
  "category": "Factual Recall",
  "question": "What are the 14 control families in NIST SP 800-171?",
  "expected_keywords": [...]
}

Coverage (Limited)

This preview covers a small sample across:

Factual recall (control family identification)
Basic control lookups
Framework identification (CMMC Levels 1, 2, 3)
Hallucination traps (Level 4/5 refusal)

It does NOT cover:

Document generation (POA&M, SSP drafting)
Gap analysis scenarios
Cross-framework mapping
Assessment guidance
Professional judgment testing
Consistency testing
Regulatory update awareness

For full coverage, use v3 comprehensive.

Citation

@dataset{memoriant-cmmc-benchmark-v1-preview-2026-q2,
  author = {Maine, Nathan},
  title = {CMMC Benchmark v1 Preview Q2 2026},
  year = {2026},
  month = {April},
  publisher = {Memoriant, Inc.},
  url = {https://huggingface.co/datasets/memoriant/cmmc-benchmark-v1-preview-2026-q2},
  note = {Preview sample. For real evaluation use v3 comprehensive.}
}

Related Memoriant Assets

memoriant/cmmc-training-data-2026-q2 — Training corpus
memoriant/cmmc-benchmark-v2-spotcheck-2026-q2 — Spot check tier
memoriant/cmmc-benchmark-v3-comprehensive-2026-q2 — The authoritative evaluation standard

Changelog

2026-q2 (Current)

Rebranded with quarterly versioning and "preview" tier labeling
Converted to auto-gated access with responsible use acknowledgement
Added AI safety disclaimer
Added expiration and refresh messaging
Repositioned as methodology sample, not evaluation tool
Preserved history from predecessor (memoriant/cmmc-compliance-benchmark)

Future Releases

2026-q3 (July 1, 2026)
2026-q4 (October 1, 2026)

Contact

Memoriant, Inc.

Website: memoriant.ai
HuggingFace: memoriant

Published by Memoriant, Inc. — For real compliance AI evaluation, use v3 comprehensive.

Downloads last month: 12

Total file size:

23.2 kB