The act of taking information directly from an existing classified source and stating it word-for-word in a new document is an example of: Copying, Not Paraphrasing

The act of taking information directly from an existing classified source and stating it word-for-word in a new document is an example of copying — specifically, verbatim copying, also called direct copying or extraction. In classification and information-security training, this behavior is the textbook example of a derivative classification error when the new document fails to carry forward the original classification markings. The boundary matters: word-for-word reuse of classified material is allowed only through authorized derivative classification, with proper markings, source attribution, and handling controls.

That's the answer most people are looking for, and it's the one that matters in practice. But there's a second reading, and it's the one that trips people up on exams. The same act, described outside a security context, is cited as an example of plagiarism. The word-for-word part is identical; what changes is the frame around it. One frame is about national security markings, the other about academic and professional attribution.

Consider a cleared analyst who pulls three paragraphs from a Secret//NOFORN assessment into a new briefing slide deck. If the analyst marks the deck Secret//NOFORN and cites the source document in the classification authority block, that's proper derivative classification. If the analyst pastes the same three paragraphs into an unclassified white paper without markings, that's a spillage. Same words, radically different consequence.

The consequences are not symmetrical, and that's the point people miss. The classification reading carries criminal and administrative exposure under executive order and agency regulation. The plagiarism reading carries professional and academic consequences. Both can apply to the same act if the source was classified and the reuse was unattributed.

One more boundary before we go further. Copying is distinct from paraphrasing, and the distinction is tested constantly. Paraphrasing restates the idea in new words; copying reproduces the expression. The classification system cares about the information's substance, so a paraphrase of classified content is still classified. The plagiarism system cares about the expression, so a paraphrase with attribution is generally fine. Keep those two axes separate in your head and most exam questions resolve themselves.

Why Word-for-Word Reuse Triggers Derivative Classification Rules

Derivative classification is the act of incorporating, paraphrasing, restating, or generating in new form information that is already classified, and marking the new material consistent with the source. The definition comes from the Information Security Oversight Office's implementing directive for Executive Order 13526, the order that has governed classified national security information since December 2009. Note the phrase "incorporating, paraphrasing, restating" — copying is explicitly inside the definition, not an exception to it.

That means the act in the question isn't forbidden. It's regulated. The moment you lift a sentence from a classified source into a new document, you have created a derivatively classified document and you inherit the obligation to mark it. The original classifier — the "original classification authority" — made the judgment that the information needs protection. You are not re-making that judgment; you are carrying it forward.

The marking obligations are specific. A derivatively classified document needs the overall classification, portion markings where required, and a "Derived From" line identifying the source document, plus a "Classification Reason" code in many agency implementations. The Defense Counterintelligence and Security Agency's marking guides walk through the exact placement. Miss the "Derived From" line and the document is improperly marked even if the banner at top and bottom is correct.

Here's where people get caught. The most common derivative classification error isn't malicious copying — it's copying without carrying the markings forward. An analyst builds a new document from three classified sources, marks it at the highest level, and forgets that one paragraph came from a source with a more restrictive control marking like NOFORN or a special access caveat. The banner says Secret; the paragraph needs Secret//NOFORN. That mismatch is a reportable error in most agency self-inspection programs.

The failure case is worth spelling out because it's the one training rarely dramatizes. A command's self-inspection finds a derivatively classified slide with the right banner but no source citation. The slide gets traced, the source turns out to be a compartmented program document, and the slide now sits in a briefing folder distributed to 40 people who lack the caveat access. The remediation is a recall, a re-brief, and an entry in the command's classification incident log. Nobody stole anything. Somebody copied and didn't finish the paperwork.

Copying Versus Paraphrasing: What the Classification System Actually Cares About

The classification system's test is whether the information is classified, not whether the words are new. This is the single most important distinction for anyone preparing for a derivative classification exam, and it's the one most study guides underweight. You can paraphrase a classified paragraph completely and produce an unmarked document that is just as much a spillage as a verbatim paste.

The practical rule: if the underlying fact, judgment, or assessment came from a classified source, the new document needs derivative markings regardless of how different the prose looks. Rewriting "the program is behind schedule by nine months" as "schedule slippage of roughly three quarters" changes nothing about the classification status. The information is the same information.

Where paraphrasing does matter is in detection and in attribution practice. Verbatim copying is trivially detectable by string-matching tools that agencies run on draft documents; paraphrased copying is not. That asymmetry explains why so much training focuses on the verbatim case — it's the easy one to catch, and it's the one that shows up in misconduct cases because the copying was lazy.

A useful mental model has two independent axes. Axis one: is the source classified? Axis two: is the reuse verbatim or paraphrased? All four combinations exist, and three of the four require derivative markings if the source was classified. Only the unclassified-source combination is free of the classification obligation, and even then the plagiarism obligation may attach.

The edge scenario that confuses people: quoting a classified source inside a document that is itself properly marked. That's fine, and it's routine in intelligence products that quote source reporting. The quote marks matter for a different reason — they tell the reader that the wording is the source's, which is an analytic tradecraft convention, not a classification requirement. Keep the two conventions separate and you'll mark documents correctly.

The Same Act Under Academic and Professional Rules: Plagiarism

Outside the security frame, word-for-word reuse without quotation and citation is plagiarism, and the standard definitions are blunt about it. The American Psychological Association's publication manual treats verbatim reuse of seven or more words from a source as requiring quotation marks and a page citation; the Modern Language Association's handbook uses similar thresholds in practice guidance. Those numbers are conventions, not law, but they're the ones graders and editors apply.

The classification and plagiarism frames answer different questions. Classification asks: who is authorized to see this? Plagiarism asks: who gets credit for this expression? A cleared analyst can be perfectly compliant on classification and still plagiarize a colleague's unclassified analysis memo by pasting it into a report without attribution. The colleague's remedy is professional, not criminal.

Institutions vary in how they handle the overlap. University research-misconduct policies typically define plagiarism as "the appropriation of another person's ideas, processes, results, or words without giving appropriate credit," a formulation that appears in the U.S. Office of Research Integrity's federal policy on research misconduct. That definition covers paraphrased appropriation too, which is why research-integrity training spends as much time on close paraphrase as on verbatim copying.

The practical difference shows up in the paper trail. Plagiarism findings live in institutional files and can end a career; classification errors live in security incident systems and can end a clearance. A single act of copying a classified, unpublished colleague's memo into a new document can generate both, in two separate proceedings, with two separate sets of decision-makers. That's not double jeopardy in any legal sense — the interests protected are different.

Where Verbatim Copying Is Legitimate: Quotation, Statutes, and Standard Language

Copying is not a sin in itself, and the exceptions are worth knowing because they show how the rules actually work. Direct quotation with attribution is the obvious one. Statutes, regulations, and court holdings are routinely quoted verbatim because paraphrase would create ambiguity about what the binding text says — a contract clause reproduced word-for-word is more useful than a summary.

Standard language is another legitimate case. Classification guides themselves contain standard derivative marking blocks that are meant to be copied exactly. Safety warnings, required disclosures, and regulatory boilerplate exist precisely so that people copy them without alteration. An agency that paraphrased its own mandatory markings would create marking inconsistency, which is the problem the standard language was designed to solve.

The boundary between legitimate and illegitimate copying is authorization plus attribution. A quotation is authorized by the source's public availability and attributed by the citation. A derivatively classified paragraph is authorized by the analyst's clearance and attributed by the "Derived From" line. An unattributed paste is neither. The act is identical; the surrounding controls are what make it acceptable.

A worked example makes the arithmetic concrete. Suppose an analyst drafting a 20-page assessment pulls one paragraph of roughly 120 words from a classified source into each of eight pages. That's 960 words of verbatim reuse out of a document whose total body text runs about 9,000 words — roughly 11 percent of the document is copied. Every one of those eight paragraphs needs a portion marking consistent with its source, and the document needs a "Derived From" line listing each source. Miss two and the error rate is 25 percent of the copied paragraphs, which is the kind of number that turns a routine self-inspection into a findings memo.

What the Two Regulatory Frameworks Require in Practice

The classification framework runs through Executive Order 13526 and its implementing directive, 32 CFR Part 2001, issued by the Information Security Oversight Office. That directive sets the training requirement: within two years of initial classification authority and annually thereafter, persons who derivatively classify must receive training covering classification levels, markings, prohibitions, and sanctions. The training is not optional, and agencies track completion.

The sanctions language is where the framework gets teeth. The executive order directs that officers and employees who knowingly, willfully, or negligently disclose classified information to unauthorized persons be subject to sanctions, including reprimand, suspension, loss of clearance, or removal. Negligence is in that list. An unmarked verbatim copy that ends up in the wrong hands is a negligence case even without intent.

The academic framework runs through institutional policy and, for federally funded research, the Office of Research Integrity's misconduct regulations at 42 CFR Part 93. That regulation's definition of plagiarism excludes honest error and differences of opinion, which is a meaningful carve-out: a citation format mistake is not research misconduct, but wholesale appropriation of another's text is.

Compare the two frameworks on one dimension that matters for practitioners. Classification violations are adjudicated inside the security apparatus, with clearance eligibility as the currency. Plagiarism findings are adjudicated inside the institution, with employment or publication status as the currency. A person can lose a clearance without losing a job, or vice versa, and the two decisions proceed on different evidence standards. Knowing which tribunal you're facing determines what you do first when a copying problem surfaces.

Marking, Attribution, and Handling: A Compliance Sequence

When you're about to reuse classified content word-for-word, the sequence below is the order that keeps a document compliant from draft to final. It's drawn from the marking conventions in agency derivative classification guides, which follow the ISOO directive's structure. Run it every time, not just when the source feels sensitive.

1: Identify the source document's overall classification, portion markings, and any control markings such as NOFORN, REL TO, or compartment caveats. 2: Confirm your own access to every marking on the source before you copy a single sentence. 3: Mark the copied portion at the source's level, including all control markings, not just the top-line classification. 4: Add a "Derived From" line naming each source document by date and subject, and a classification reason code where your agency requires one. 5: Apply overall marking at the highest level present in the document, then run a second pass comparing each copied portion against its source marking.

If step two fails — you lack access to a caveat on the source — stop and route the question to your security office before drafting further. If step five turns up a mismatch, fix the portion marking and log the correction per your command's classification incident procedure. The failure-handling close matters more than the sequence itself, because the sequence is in every training slide and the recovery path is not.

A Side-by-Side Comparison of the Three Readings

The same sentence about copying classified material word-for-word can be graded under three different rulebooks, and the outcomes diverge sharply. The table below compares them on the dimensions that decide cases: what triggers the finding, who adjudicates, and what the realistic consequence is. None of the figures are invented — they come from the frameworks named in the rows.

The verdict depends on who is asking. A security officer cares about row one, a dissertation committee about row two, and a publisher's counsel about row three. If you're answering an exam question, the grading key almost always wants row one when the word "classified" appears in the stem, and row two when it doesn't.

The row most people never consider is the third. Copyright and classification are independent: a classified document can still be a copyrighted work, and its unauthorized public reproduction can infringe even if the classification handling was flawless. Government works prepared by federal employees are generally not subject to copyright in the U.S., but contractor-prepared classified documents can be, which is why some derivative products carry contractor rights legends alongside their security markings.

How Exams Frame the Question and How to Answer

Derivative classification exams, including the Defense Counterintelligence and Security Agency's widely used course, phrase this question in a handful of stable patterns. The most common stem is the one in this article's title, and the expected answer is "derivative classification" when the context is a cleared environment and "copying" when the question asks for the plain-language behavior. Read the answer options before committing — the distractor set usually includes "paraphrasing," "original classification," and "declassification," and each is wrong for a specific reason.

Original classification is the trap option worth understanding. Original classification is the initial determination that information requires protection in the interest of national security, and it can only be made by an official with delegated original classification authority. Copying from an existing source can never be original classification, because the judgment was already made upstream. If an exam option says "original classification" for a copying scenario, it is wrong by definition.

A second stable pattern asks what the copier's marking obligation is, and the correct answer is that the new document must be marked consistent with the source. A third asks about paraphrasing, and the correct answer is that paraphrasing does not remove the classification. A fourth, less common, asks about the sanction for negligent handling, and the answer traces to the executive order's sanctions clause.

The failure scenario to rehearse is the one where you cannot determine the source's markings. Suppose you inherit a draft paragraph with no source citation and an overall Secret banner. You cannot verify whether the paragraph carries a caveat, and copying it forward risks spreading a control marking you can't see. The correct move is to treat the paragraph as unusable until the source is identified, and to flag the draft to whoever handed it to you. Exams rarely test this, but real security offices see it constantly.

Common Misconceptions That Cost People on Tests and at Work

The first misconception is that copying "just the facts" escapes classification. It doesn't, and this is the misconception that shows up in real incidents. Facts, assessments, and judgments inherit their classification from the source; only genuinely public information loses it, and the burden is on you to establish that the information is public.

The second is that quotation marks cure a marking problem. They don't. Quote marks are an attribution convention. A quoted classified sentence still needs a portion marking, and a quoted unclassified sentence still needs a citation. The two conventions stack; neither substitutes for the other.

The third is that derivative classification is a lesser category than original classification, and therefore lower stakes. The training data says otherwise. Agency self-inspection reports consistently find derivative classification errors outnumber original classification errors, because far more people do derivative work. The volume is where the risk lives.

The fourth, and the one that produces the worst outcomes, is that intent determines whether a copying incident is reportable. It doesn't. Negligent handling is sanctionable under the executive order, and most agency incident-reporting procedures require reporting regardless of intent. A well-meaning analyst who pastes a paragraph into the wrong draft still generates a reportable incident, and the report is what triggers the review — not the motive.

FAQ

What is the difference between derivative classification and declassification?

Declassification is the removal of classification status by an authorized official, and it is the opposite direction of travel from derivative classification. A derivative classifier carries existing classification forward; a declassification authority removes it after review. Copying word-for-word never declassifies anything — the copied text retains its status until an authorized declassification action occurs.

Can I copy a classified paragraph into an unclassified document if I paraphrase it?

No. Paraphrasing does not strip classification, and an unclassified document containing paraphrased classified content is a spillage just like a verbatim paste. The correct path is to route the draft through your security office for a derivative classification review before it goes anywhere. Rewriting the prose is not a substitute for that review.

Does a quotation with citation satisfy the marking requirement?

No, quotation marks and a citation satisfy the attribution convention but not the classification marking requirement. A quoted classified sentence still needs a portion marking consistent with its source and a "Derived From" line in the document. The two conventions address different obligations and both apply when the source is classified.

How often must derivative classifiers be trained?

The ISOO implementing directive requires derivative classification training within two years of initial classification authority and annually thereafter, and agencies typically run the refresher on a 12-month cycle. Completion is tracked, and lapse in training is itself a finding in many self-inspection programs. The specific course varies by agency, but the annual cadence does not.

What happens if a derivatively classified document is missing its "Derived From" line?

The document is improperly marked, which is a reportable classification error in most agency procedures. The usual remediation is to identify the source, correct the marking, recall and reissue any distributed copies, and log the correction. The missing line itself doesn't change the document's classification level — the content does — but the error triggers the review.

Is copying classified material always a security violation?

No. Verbatim reuse through authorized derivative classification, with proper markings and handling controls, is routine and expected. The violation is the failure to carry markings forward, the distribution to unauthorized recipients, or the copying by someone without access to the source. The act is neutral; the controls determine the outcome.

How does copyright interact with classified documents?

They're independent frameworks, and a classified document can also be a copyrighted work. Federal-employee-prepared works are generally outside U.S. copyright, but contractor-prepared classified documents can carry copyright, which is why some derivative products include contractor rights legends. Check the source document for rights language before reproducing it outside your agency.

What's the single most common derivative classification error?

Missing portion markings on copied paragraphs, according to the pattern in agency self-inspection findings, where the overall banner is correct but individual portions don't match their sources. The second most common is the missing or incomplete "Derived From" line. Both are paperwork failures on correctly understood content, and both are fixable before distribution.

This article is general educational information about classification marking and attribution conventions, not legal advice or an official interpretation of any agency's classification guidance. Classification rules, agency procedures, and academic integrity policies vary and change; confirm the governing requirements with your security office, institutional integrity office, or a qualified professional in your jurisdiction. If you're facing an active classification incident or misconduct proceeding, contact the appropriate office before acting.