The statement that most accurately describes a derivative classifier's responsibility is that the classifier must review a source classified document for classification issues and determine whether any portion of it warrants derivative classification, then apply the correct markings to the derivative work. A derivative classifier does not create original classification decisions; that authority belongs to original classification authorities. The derivative classifier's duty is to carry forward, correctly and consistently, classification decisions that already exist in an authorized source document, and to mark the newly created material accordingly.
That distinction sits at the center of every accurate statement about derivative classifier responsibilities. According to the Office of the Director of National Intelligence's marking guide, derivative classification occurs whenever information is extracted, paraphrased, restated, or generated in a new form and that new form carries classified information drawn from an authorized source (ODNI). The classifier is the person doing that extraction or restatement. The responsibilities follow from that role, not from a general security clearance or a supervisory title.
Mechanism: how derivative authority actually flows
Derivative classification authority is delegated through a chain. An original classification authority (OCA) makes the initial decision that specific information is classified, at a specific level, for a specific reason, and with a specific declassification date or event. That decision is recorded in a source document's markings, typically including the classification authority block and a reason for classification. A derivative classifier receives that decision as a given. The derivative classifier's job is to translate it into any new document, briefing, database entry, or summary.
The core mechanism is a one-way transfer of judgment. The derivative classifier cannot upgrade or downgrade the level unless the source authorizes it, cannot invent a classification reason, and cannot extend a declassification date without authority. Each of those actions would amount to an original classification decision, which the derivative classifier is not empowered to make. When a derivative classifier encounters information that is not covered by any source, the correct action is to refer it upward to an OCA, not to classify it personally.
A second mechanism matters for accuracy. Derivation is not limited to copying text. Restating a fact in different words, aggregating unclassified details into a classified conclusion, or generating a chart from classified data all count as derivative classification if the resulting product reveals classified information. This is why the responsibility statement has to include marking the new product correctly, not merely reviewing the old one.
Quantified context: the scale behind the rules
The derivative classifier role exists at massive scale. Executive Order 13526, the foundational US classification order, establishes the categories and levels that derivative classifiers apply. Its companion implementing directive, 32 CFR Part 2001, defines the training and review structure.
Training thresholds are concrete. Under 32 CFR 2001.71, personnel with derivative classification authority must receive training before exercising that authority and must be retrained annually. The same part requires that derivative classification decisions be reviewed, and requires agencies to maintain a self-inspection program. Those intervals are not suggestions; annual retraining is the compliance baseline for the derivative classification function.
The volume of derivative classification far exceeds original classification. Most classified documents created across the federal government are derivatives of a smaller set of original decisions. That asymmetry is the reason the role is rule-driven: a single original decision may propagate into hundreds of derivative products, so an error at the top propagates at scale.
Procedural numbers also anchor the role. Derivative classifiers must mark each portion with the classification level and, where required, the category. They must include the "Classified By" line identifying the source, the "Derived From" line citing the source document, a declassification instruction, and a reason for classification. Missing any of these is a marking error even when the classification level itself is correct.
Comparison: derivative classifier versus adjacent roles
| Attribute | Original classifier (OCA) | Derivative classifier | Reviewer/Custodian |
|---|---|---|---|
| Makes new classification decisions | Yes, within delegated authority | No | No |
| Applies markings to new products | Only if also deriving | Yes, required | Verifies, does not create |
| Can upgrade or downgrade level | Yes, with authority | Only per authorized source | No |
| Sets declassification date | Yes | Carries forward from source | Checks for completeness |
| Training cadence | Per agency policy | Annual, per 32 CFR 2001.71 | Varies by program |
| Handles uncovered information | Classifies or refers | Refers to OCA | Escalates |
The table's central row is the upgrade/downgrade line. That is the boundary that most accurately defines the role. A derivative classifier who changes a classification level without source authority has exceeded the role, regardless of how reasonable the change seems.
Edge cases where the role gets tested
Several recurring situations separate a correct derivative classifier statement from a merely plausible one.
Compilation effects: A new document combining individually unclassified facts may reveal a classified relationship, but the derivative classifier still needs a source that marks that relationship as classified. Without it, the correct step is referral, not independent classification. Partial source texts: If a source document is classified in its entirety but the excerpt is unclassified, the classifier must assess whether the excerpt alone reveals classified information. The answer depends on the source's portion markings, not on the document's overall stamp. Foreign government information: This carries its own handling rules and may require the derivative classifier to preserve the foreign marking, not replace it. Working papers: Drafts and notes used to prepare a finished product may themselves be classified if they contain classified content, so the responsibility extends to intermediate artifacts. Legacy documents: Older sources may have markings that predate current rules. Applying a current-format marking to a legacy source without checking the source's instructions creates inconsistencies. Multi-source products: The classifier must list all sources in the derived-from line and must resolve conflicting instructions by choosing the most restrictive, not by averaging.
The professional caveat here is that most derivative classification errors are marking errors, not judgment errors. Audits frequently find level markings applied correctly but derived-from lines missing or declassification instructions copied incorrectly. That pattern suggests the role is best described as precision work, not analysis work. A plain-language way to say it: the derivative classifier is a careful copyist of decisions, not a decision-maker.
Misconceptions that produce wrong answers
The most common misconception is that a derivative classifier can classify newly created information. That misconception produces an inaccurate statement: a derivative classifier may not assign a classification level to information not already classified in an authorized source. Only an OCA may do that.
A second misconception treats clearance level as authority. Holding a top secret clearance does not confer derivative classification authority; the authority comes from being designated to perform the function and from the training that accompanies the designation.
A third misconception confuses classification with controls. Handling caveats, dissemination controls, and need-to-know rules are applied by the derivative classifier, but they are not classification levels. A statement that the derivative classifier determines need-to-know is describing an access control function, not the core classification responsibility.
A fourth misconception assumes the classifier may paraphrase freely. Paraphrasing does not remove classification. Restating classified content in new words creates a derivative, and the new wording must be reviewed for whether it still reveals the classified substance.
A fifth misconception assumes a single source is always sufficient. Multi-source derivatives require reconciling multiple instruction sets, and the classifier must carry forward the most restrictive declassification instruction among the sources unless an OCA directs otherwise.
Timeline: how a derivative product is handled
| Stage | Action | Typical owner |
|---|---|---|
| 1. Source selection | Identify an authorized source with valid markings | Derivative classifier |
| 2. Content mapping | Determine which portions of the new work derive from the source | Derivative classifier |
| 3. Level and category assignment | Apply source levels to each portion | Derivative classifier |
| 4. Marking | Add classified-by, derived-from, declassification, and reason lines | Derivative classifier |
| 5. Review | Check markings for consistency and completeness | Reviewer or supervisor |
| 6. Distribution controls | Apply handling caveats and need-to-know | Custodian or classifier |
| 7. Annual retraining | Refresh derivative authority | Designated official |
The sequence matters because errors caught at stage 5 are cheap. Errors that survive to distribution propagate into every downstream copy, and correcting them later requires tracking every recipient. That is why the marking step is the one derivative classifiers are most often evaluated on.
Practical actions: what to do when the question is ambiguous
When reviewing a candidate statement about derivative classifier responsibilities, apply a short test.
Confirm the statement describes carrying forward an existing decision rather than creating a new one. Confirm it mentions marking the new product, not just reviewing the source. Confirm it preserves the source's level and declassification instruction. Flag any statement that implies independent authority to classify, declassify, or change levels. Check whether the statement describes an access control as if it were classification. For multi-source products, check that the most restrictive instruction is being carried forward.
For someone actually performing the role, the operating thresholds are these. Complete training before first use of derivative authority and annually thereafter. Verify the derived-from citation on every product. Treat any uncovered information as a referral case. Record which source authorized each marking decision so a reviewer can trace it in one step.
The single most useful habit is a two-question check applied to every statement about the role. First, does the statement describe applying an existing decision or making a new one? Second, does it describe marking the output? Statements that pass both describe the responsibility accurately.
Limitations and what the rules cannot settle
Written rules cannot resolve every real case. Classification decisions can be genuinely contested, sources can conflict, and legacy markings can be internally inconsistent. In those situations the derivative classifier's accurate responsibility is still referral and documentation, not unilateral resolution. Agency-specific directives also vary, so a statement that is accurate at one agency may need qualification at another.
This article is general educational information about classification procedures, not legal advice or an agency directive. Derivative classification rules vary by jurisdiction and agency, and specific questions about authority, marking requirements, or compliance should go to a qualified security or legal professional within your organization. For immediate security incidents, follow your agency's reporting procedures.