Why the origin of information does not determine whether it is Controlled Unclassified Information
The Myth of Derivative CUI
There is a widely repeated assertion within the cybersecurity and defense contracting communities that anything derived from Controlled Unclassified Information (CUI) must itself be CUI. The argument usually goes something like this: If you receive a document containing CUI and use that information to create another document, the new document is automatically CUI. The same reasoning is frequently extended to spreadsheets, reports, emails, engineering analyses, and other information products. This concept is commonly referred to as derivative CUI.
There is a fundamental problem with this assertion. There is no general regulatory requirement that information automatically becomes CUI simply because it was derived from CUI. In fact, the National Archives and Records Administration (NARA), which serves as the CUI Executive Agent, addressed this precise issue when it promulgated the CUI implementing regulation in 2016. The designation depends on the information itself, not merely where it came from. Understanding this distinction has significant implications for government contractors, particularly those attempting to establish appropriate CUI boundaries, implement NIST SP 800-171, and comply with the Cybersecurity Maturity Model Certification (CMMC) program.
CUI Is Information, Not a Document
Under 32 CFR § 2002.4, CUI is information that the government creates or possesses, or that an entity creates or possesses for or on behalf of the government, that a law, regulation, or government-wide policy requires or permits an agency to safeguard or control. Notice that the definition concerns information. It does not say that every document, file, or other object created using another document marked CUI automatically inherits that designation.
For purposes of this discussion, I propose a useful term: information object. An information object is any discrete artifact, record, representation, or collection that contains or communicates information, regardless of its physical or electronic form. An information object might be an engineering drawing, spreadsheet, email, database record, computer file, presentation, or handwritten note. It may contain CUI, non-CUI, or a combination of both. This is the thing that holds the CUI. So I make a new document, spreadsheet, cad drawing, whatever, that is an information object. Any time we create a new information object, and information container if you will, we must be mindful of the information that goes in there much more so than where it came from.
When an authorized holder creates a new information object, the question is not whether the object was derived from another object containing CUI. The question is whether the information contained in the new object qualifies as CUI under the applicable legal authority. That determination should be made regardless of whether the information originated in a CUI-marked document, an unmarked document, or entirely new work. This does not mean every new file requires a formal, documented determination. It means that information must be properly identified, marked, and handled when applicable CUI requirements call for it.
NARA Already Answered This Question in 2016
When NARA published the final CUI implementing regulation on September 14, 2016, it included responses to public comments concerning the proposed regulation. One commenter specifically raised the question of information derived from CUI.
“One comment requested an explanation of the status of information derived from CUI, and another suggested we add a requirement to mark the designating and disseminating agencies on all CUI.” Source
“Regarding the questions about derived CUI, the bottom line is that certain types of information qualify as CUI. If an item of information qualifies as CUI, it doesn't matter whether it is in some way also derived from another item of information that qualifies as CUI, and it should be marked as CUI either way. Its status as CUI depends upon the information itself and whether it meets the requirements in a law, regulation, or Government-wide policy that establish it as needing controls on safeguarding or disseminating. A document containing CUI that is derived from another document that contains CUI would also be CUI—because it contains controlled information, not simply because it is derived from a document that contains CUI. It is possible the original document contains both CUI and non-CUI and the derived document could therefore contain only information derived from the non-CUI portions of the original document. In such a case, the derived document would not become CUI simply because the information was derived from a CUI document.” Source
In this discussion of the final rule, NARA explains that the status of information derived from a CUI document depends on the information itself, rather than merely its origin. A source document might contain both CUI and information that does not qualify as CUI. A subsequent document containing only the uncontrolled information does not automatically become CUI simply because the source document was marked CUI. Conversely, if the subsequent document reproduces, summarizes, or otherwise communicates information that meets the applicable CUI criteria, that information remains subject to the appropriate controls.
The determining factor is the information contained in the new object, not the designation of the source object. This is consistent with the explanation provided by the agency responsible for administering the CUI Program.
What 32 CFR Part 2002 Actually Requires
The CUI implementing regulation establishes a framework for identifying, safeguarding, marking, disseminating, and decontrolling information. It does not establish a general rule of automatic inheritance for every product derived from CUI. Section 2002.1 establishes that CUI controls must be grounded in applicable law, regulation, or government-wide policy. Section 2002.20 establishes requirements for identifying and marking CUI, including the responsibilities of agencies and authorized holders. The regulation recognizes that different portions of an information object can have different control statuses. A document may contain CUI alongside information that is not controlled. Extracting controlled information and placing it in a new information object does not eliminate applicable safeguards.
Example 1 — Copying controlled technical information: A contractor copies controlled engineering specifications into a spreadsheet. The spreadsheet contains CUI because it contains controlled technical information, not merely because it is related to the drawing.
Example 2 — Creating an administrative schedule: The contractor uses the drawing to plan work and creates a schedule of assignments, meetings, and milestones. Assuming it contains no information qualifying under an applicable CUI category, the schedule is not automatically CUI. This schedule references applicable Mil Standards. Although Mil Standards are technical, they are also generally publicly available and not CUI.
Example 3 — Creating new technical analysis: A contractor analyzes controlled specifications and writes a report revealing system performance characteristics. Even without reproducing the source text, the conclusions may independently qualify as controlled technical information.
In all three examples, the same principle applies: Evaluate the information, not merely its ancestry.
DFARS 252.204-7012 Reinforces This Principle
DFARS 252.204-7012, Safeguarding Covered Defense Information and Cyber Incident Reporting, establishes safeguarding and reporting requirements for covered defense information (CDI). The CDI definition points to the CUI categories as its foundation.
Paragraph (m)(1), addressing subcontractors, requires a contractor to determine whether information needed for subcontractor performance retains its identity as covered defense information. The regulation expressly contemplates a determination about whether information passed to a subcontractor retains its CDI status. If mere origin automatically determined designation, that inquiry would be largely unnecessary.
The clause does not give contractors general authority to decontrol CUI, nor does it provide a comprehensive framework for newly generated information. Decontrol is a separate concept not to be confused with derivation. Decontrol has separate legal authorities generally retained by the federal government. This clause is specific subject is subcontractor flowdown, and it reinforces the importance of evaluating information in a new information object rather than a blanket derivation rule.
Even Classified Information Does Not Work That Way
We can also look to how this works with more sensitive government information: classified. The federal government maintains a formal derivative-classification system for classified national security information under Executive Order 13526. Individuals creating new classified products must apply the applicable classification guidance and restrictions. However, this does not mean that every information product derived from Top Secret information must itself be classified Top Secret. During my 21-year career as a Navy Cryptologist, I worked extensively with classified intelligence, including preparing information for audiences with different classification and dissemination requirements. It was routine to develop lower-classification products from higher-classification source material through sanitization. Top Secret intelligence might be used to produce a Secret-level report, or even an unclassified product, provided the resulting information was appropriately evaluated and approved for release.
Similarly, information originally restricted from foreign dissemination might be used to produce a version releasable to specified foreign partners, subject to applicable disclosure authorities. Sanitization does not confer unrestricted authority to downgrade, declassify, or release information. Appropriate classification, declassification, and foreign-disclosure authorities must be respected. Even the classified-information system where the information is much more sensitive, does not impose a universal requirement that every resulting product inherit the classification of its source material.
Why This Matters for CMMC
The mistaken belief that everything derived from CUI automatically becomes CUI can have significant operational and financial consequences for defense contractors. An engineering organization may receive controlled technical information and use it in manufacturing, quality assurance, procurement, scheduling, project management, and financial planning. If every information object generated during those activities is presumed to be CUI, administrative spreadsheets, scheduling systems, internal communications, and business applications may be unnecessarily treated as CUI assets. That can drive broader CMMC assessment boundaries, additional technical safeguards, and substantial implementation costs.
Of course, systems that actually process, store, or transmit CUI must be appropriately protected. Systems providing security protection for CUI assets may also be in scope under applicable CMMC scoping rules. The point is not that contractors can avoid safeguards by transforming information into another format. CUI scoping should begin with an accurate determination of what information is actually CUI.
Contractors Cannot Simply Declare CUI Uncontrolled
Rejecting automatic derivative CUI does not mean contractors have unrestricted authority to remove CUI markings or disregard government designations. The CUI Program establishes responsibilities for federal agencies and authorized holders, including procedures for correcting or challenging improper designations. Contractors must comply with contract requirements and seek clarification from the responsible prime contractor or agency when necessary. An agency may provide contract-specific instructions for identifying and handling newly created information. Those instructions must be considered. Proper identification and handling responsibilities are not the same as unilateral authority to decontrol information.
We Need to Stop Treating Derivation as Designation
The cybersecurity community frequently uses shorthand to explain complicated regulatory requirements. Sometimes that shorthand becomes so widely repeated that people begin treating it as if it were the regulation itself. This appears to be ongoing with derivative CUI. The assertion that anything derived from CUI must automatically be CUI is easy to remember and simple to teach. It is also an inaccurate statement of the general regulatory framework.
Information must be evaluated against applicable CUI categories and governing authorities. Authorized holders must identify, mark, safeguard, and disseminate information according to their responsibilities. Contractors must follow applicable contractual requirements and agency instructions. NARA addressed the issue in 2016. DFARS 252.204-7012 reinforces the importance of determining whether information retains its controlled identity. Even the classified-information system recognizes properly authorized derivative products can have different classification and dissemination controls from their sources.
CUI is a designation based on the information and the legal authority governing it. It is not a hereditary characteristic of documents, files, or other information objects. The next time someone asserts that anything derived from CUI is automatically CUI, ask a simple question: Where, exactly, does the regulation say that?
Regulatory References
- 32 CFR Part 2002 — Controlled Unclassified Information, particularly §§ 2002.1, 2002.4, and 2002.20. Source
- 81 FR 63324 — Controlled Unclassified Information, Final Rule, September 14, 2016; discussion beginning at 63334. Source
- DFARS 252.204-7012 — Safeguarding Covered Defense Information and Cyber Incident Reporting, particularly paragraphs (a) and (m). Source
- DoDI 5200.48 — Controlled Unclassified Information. Source
- Executive Order 13526 — Classified National Security Information, particularly § 2.1. Source