Information May Be Classified If It Is Owned by the U.S. Government: The Embarrassment Standard Explained

Information owned by the U.S. government may be classified even when its unauthorized disclosure would only embarrass the government and would not harm national security, and even when it concerns failed scientific research. That is the honest reading of the governing executive order, and it's a point officials rarely volunteer. Classification is not limited to genuine secrets of state. It is an administrative control the government applies to its own records, and the embarrassment category sits alongside national security categories rather than beneath them as a lesser tier that somehow needs security harm to justify it.

What follows separates the myth from the executive order text, the statutory backdrop, and the practical consequences for researchers, journalists, and federal employees who encounter a failed experiment inside a government file.

The executive order text that makes embarrassment a classification category

The controlling authority is Executive Order 13526, "Classified National Security Information," signed by President Obama on December 29, 2009, which replaced the Bush-era Order 13292 and remains the framework in force as of this writing unless a successor order has superseded it. Section 1.4 lists the categories of information that may be classified, and the wording matters. The order frames classification around "damage to the national security," then enumerates specific classes that qualify.

One of those enumerated classes, at 1.4(c), covers "intelligence sources and methods." Another, at 1.4(d), covers "foreign relations or foreign activities of the United States, including confidential sources." The embarrassment language appears in the context of foreign relations: information whose disclosure could "damage the foreign relations of the United States" or "reveal a confidential source." The phrase "embarrass the government" appears in the order's discussion of what classification must not be used for — a prohibition, not an authorization.

Here is the practical crux. The order's prohibition language says classification shall not be used to "conceal violations of law, inefficiency, or administrative error" or to "prevent embarrassment to a person, organization, or agency." That prohibition is widely cited as if it settles the matter. It does not. The prohibition binds the original classifying authority's stated rationale, but agencies retain broad discretion to characterize the same information as fitting an authorized category, and courts defer heavily to agency classification determinations. The gap between the stated rule and the operational reality is where the keyword's premise lives.

A reader who wants to verify the current text should consult the executive order as published in the Federal Register or the National Archives' controlled unclassified information guidance, since amendments and successor orders change section numbering. The 2009 order's structure is the reference point most secondary sources describe.

What "owned by the U.S. government" actually covers

Government ownership is broader than most people assume. Under the Federal Records Act and Title 44 of the U.S. Code, a federal record is any documentary material made or received by a federal agency in connection with the transaction of public business, regardless of who physically holds it. That includes contractor deliverables, grantee reports, and in many cases the raw data behind a federally funded study.

Failed scientific research is a clean example of where ownership and sensitivity collide. A federal agency that funds a study owns the resulting records even when the experiment produced null results, and even when the principal investigator is at a university. The Bayh-Dole Act of 1980 gives contractors and grantees rights to elect title in inventions, but it does not transfer ownership of the underlying research records. Those records stay federal records subject to the agency's control.

Consider the practical path. A national laboratory runs an experiment on a proposed materials process. The process fails. The lab writes an internal report documenting the failure, the wasted expenditure, and the flawed assumptions. That report is a federal record. It is also, under 1.4(c) or 1.4(d) depending on the program, plausibly classifiable if the agency can tie it to intelligence methods or foreign collaboration. The failure itself does not immunize the document.

The critical distinction: the government owns the record, but ownership alone does not classify it. Classification requires a derivative classification decision or an original classification authority's determination that the information fits an authorized category and that unauthorized disclosure could reasonably be expected to cause the specified damage. The embarrassment premise in the keyword reflects what happens when that damage assessment is stretched.

Why failed research is a special case for classification

Failed research is unusually vulnerable to classification for three reasons, and none of them is about national security in the strict sense. First, failure documentation often contains the most candid internal criticism, which is exactly what the embarrassment prohibition was written to protect from concealment — and exactly what agencies have the strongest incentive to restrict. Second, failed research is cheap to classify retroactively because it rarely circulates widely, so the classification decision faces little internal friction.

Third, and most concretely, failed research frequently sits inside programs whose very existence is classified. A failed component of a classified program inherits the program's classification level even if the failure itself reveals nothing sensitive. This is the "derivative classification" mechanism under section 2.1 of E.O. 13526. A derivative classifier — often a trained contractor or agency employee with no original classification authority — can mark a document classified simply because it incorporates or derives from already-classified material.

The table's fourth row is the one that matters most for the keyword's premise. The order says you cannot classify to prevent embarrassment. The table's second and fifth rows show how the same failure escapes that prohibition once it is folded into a classified program or a foreign-relations rationale.

Deconstructing the keyword's claim, clause by clause

Take the keyword apart. "Information may be classified if it is owned by the U.S. government" — true as a necessary condition, though the order requires more. "Its unauthorized disclosure could embarrass the government but not harm national security" — this is the contested clause, because the order's text does not authorize embarrassment as a standalone basis. "And it concerns failed scientific research" — the clause that makes the scenario concrete and testable.

The claim is partly right and partly a misreading of how classification actually works. The order's section 1.7 prohibition is real and enforceable in principle through the Interagency Security Classification Appeals Panel, which reviews classification challenges brought by document owners and the public. ISCAP has declassified material over agency objection, but its reach is limited and its process slow. A researcher challenging a failed-research classification can expect a multi-year path through agency review, then ISCAP, with no guarantee of reversal.

What the keyword gets right is the operational outcome. In practice, agencies have classified material that is embarrassing rather than dangerous, and the prohibition has not stopped them, because the prohibition targets stated rationale and agencies supply different rationales. What the keyword gets wrong is treating that outcome as the rule rather than a documented deviation. The distinction matters for anyone deciding whether to challenge a classification.

A concrete historical reference point is the controversy over classification of information about the 2001 anthrax letter investigations, where internal assessments were withheld on grounds that mixed law enforcement sensitivity with reputational concerns. Another is the long-running dispute over whether the government's own records of failed missile defense tests were properly classified. In both cases the formal rationale was not embarrassment, even where critics argued embarrassment was the real driver.

The statutory and regulatory scaffolding around the order

Classification rests on more than the executive order. The Atomic Energy Act of 1954 created a separate "Restricted Data" category that operates outside the order's framework entirely, and failed nuclear research can fall into it. The Central Intelligence Agency Act of 1949 gives the Director of Central Intelligence authority over intelligence sources and methods that overlaps with but is not identical to the order.

FOIA is the other half of the picture. The Freedom of Information Act, 5 U.S.C. § 552, contains nine exemptions. Exemption 1 covers properly classified national security information. Exemption 4 covers trade secrets and confidential commercial information. Exemption 5 covers deliberative process material. A failed research record that is not formally classified can still be withheld under Exemptions 4 or 5, which is the quiet route for embarrassing internal findings.

The Controlled Unclassified Information program, established by Executive Order 13556 in 2010, adds a third tier. CUI is not classified, but it is restricted, and much of the federal research enterprise runs on CUI markings rather than full classification. A failed study can be marked CUI for "law enforcement" or "critical infrastructure" reasons and never appear in classified channels. Researchers who only check for classification markings miss this.

The regulatory home for CUI is 32 C.F.R. Part 2002, issued by the National Archives. That citation is worth naming because CUI decisions face far less oversight than classification decisions, and failed research frequently lands there.

How the classification process works on a failed study

Original classification authority is held by a limited set of officials designated in writing, per E.O. 13526 section 1.3. Agency heads, senior program officials, and certain military commanders hold it. A working scientist almost never does. What the scientist encounters is derivative classification — the requirement, at section 2.1, that anyone who reproduces or incorporates classified material into a new document mark it appropriately.

Derivative classification is mechanical and often overapplied. A training rule tells the derivative classifier to mark a document at the highest level of any source material it contains, or to apply the source's classification guide. If a program's security classification guide says all engineering data are Secret, then a failed component test inherits Secret. The classifier does not run a fresh damage assessment.

Declassification is the backstop, and it is slow. E.O. 13526 sets a 25-year default for automatic declassification of historically valuable records, with nine specific exemptions permitting continued classification. Failed research programs are rarely designated historically valuable, which sounds like good news for disclosure. It isn't — non-designated records can be destroyed under records schedules before any public release is considered.

The interagency appeals route runs through ISCAP, established under 32 C.F.R. Part 2001. ISCAP can reverse a classification on the merits, but its docket is small and its standards deferential. The most valuable practical advice for a challenger is to demand the classification guide and the source document that triggered the derivative marking, because the guide is where overclassification usually hides.

FOIA's role when nothing is formally classified

Here is a scenario worth walking through because it is common. A federal grant funds a clinical intervention study. The intervention fails to show effect. The agency receives the final report, notes serious design flaws, and declines to publicize the result. A journalist files a FOIA request.

The agency withholds under Exemption 5, citing deliberative process privilege, arguing the report is pre-decisional. The requester appeals. The agency releases the report but redacts the statistician's critique of the design, arguing it is internal deliberation. The released version omits the exact failure analysis. No classification was ever invoked.

This pathway explains why the keyword's premise has traction beyond classified channels. Exemption 5 and Exemption 4 do much of the work that the embarrassment prohibition was meant to prevent. The Supreme Court in NLRB v. Sears, Roebuck & Co. (1975) narrowed Exemption 5 to pre-decisional material, but agencies read "pre-decisional" broadly for research reviews that never reach a final decision.

A requester facing this can invoke the FOIA Improvement Act of 2016's foreseeable harm standard, which requires agencies to identify a specific foreseeable harm before withholding under Exemption 5. That standard has teeth, but it is litigated case by case. Naming it in the administrative appeal is the practical move.

The cost calculation behind withholding a failed study

Withholding failed research has a computable price, and the numbers are large enough to matter. Consider a hypothetical but arithmetic example built from public federal figures. A single failed multi-year research program at an agency with a $50 million annual extramural budget can consume 5 to 10 percent of that budget, or roughly $2.5 million to $5 million in one cycle. That range determines the disclosure stakes.

If the same failure is replicated by three other labs before it becomes known — a documented pattern in underpowered studies — the duplicate cost triples. A 2015 analysis in PLOS Biology by Freedman, Cockburn, and Simcoe estimated that preclinical research irreproducibility in the United States accounted for roughly $28 billion per year in wasted spending. Failed research that goes unreported feeds that number.

Disclosure of a failed study is cheap by comparison. Redacting and releasing one final report under FOIA costs an agency staff hours typically billed at an hourly rate under $100. The mismatch between a multi-million-dollar program failure and a few hundred dollars of FOIA processing is the strongest accounting argument for release.

An agency that classifies or withholds a failed study to avoid embarrassment is trading a small processing cost for a large replication cost borne by other labs. That trade never appears in the classification decision, because the deciding official does not bear the replication cost. The externality is the mechanism.

Where the embarrassment prohibition has actually been enforced

The section 1.7 prohibition is not decorative. ISCAP has cited it. In its published decisions, ISCAP has ordered declassification of information where the agency's asserted rationale amounted to reputational protection. One recurring pattern involves agency histories that described operational mistakes; ISCAP has ordered portions released over agency objection.

The Public Interest Declassification Board, established by the Public Interest Declassification Act of 2000 at 50 U.S.C. § 3161 note, has documented systematic overclassification in its reports to the President. Its recommendations have informed reforms, but the board has no authority to declassify anything itself.

Here is the boundary condition that matters. Where outside classified information is genuinely at stake — a named foreign source, a still-active method — the order's categories hold regardless of how embarrassing the underlying failure is. The embarrassment prohibition does not create a public right to disclosure. It removes embarrassment as a permitted rationale. An agency that wants to withhold can usually find another rationale, and where the other rationale is legitimate the withholding stands.

A reader challenging a specific decision should request the document's classification guide, the identifying source document, the classifier's name and date, and any applicable agency classification guide. If the stated category is foreign relations, demand the specific foreign-relations nexus. If the category is intelligence methods, demand the method. The prohibition lives in the space between a stated category and a real basis.

FAQ

Can the government really classify something just to avoid embarrassment?

No, and yes in practice. E.O. 13526 section 1.7(a) explicitly prohibits classifying information to prevent embarrassment to a person, organization, or agency, and to conceal violations of law, inefficiency, or administrative error. The catch is enforcement: the prohibition governs what a classifier may say, and agencies can usually substitute an authorized category such as foreign relations at section 1.4(d). A challenged classification is reviewed by the Interagency Security Classification Appeals Panel under 32 C.F.R. Part 2001, not by a court in the first instance, and ISCAP reversal rates are low relative to the volume of challenges. The prohibition is a real constraint on stated rationale and a weak constraint on outcomes.

Does failed scientific research get classified more often than successful research?

There is no published count, but failed research carries a structural vulnerability. It commonly sits inside programs whose classification guides apply to all engineering or all test data regardless of result, so a failed test inherits Secret marking mechanically under derivative classification rules at E.O. 13526 section 2.1. The failure also tends to generate the most candid internal criticism, which is the material the embarrassment prohibition was written to keep out of reach of concealment. The combination means failure documentation is more likely to be caught in a classification guide's net, though the underlying vulnerability is about program structure rather than the failure itself.

Is information owned by the U.S. government automatically classified?

No. Government ownership is a necessary condition for classification under E.O. 13526, but it is not sufficient. A record must also be determined by an authorized classifier to fit one of the eight categories at section 1.4, and the unauthorized disclosure must reasonably be expected to cause the specific level of damage named in the order. Most federal records are unclassified. Ownership matters because it is what brings the record inside the classification system at all; contractor reports and grantee data are federal records under 44 U.S.C. § 3301 even when the author is not a federal employee, which means they can be classified where the same material in private hands could not.

What is the difference between classified information and Controlled Unclassified Information?

Classified information requires a national security damage determination under E.O. 13526 and carries formal handling, storage, and declassification rules. Controlled Unclassified Information is established by Executive Order 13556 and 32 C.F.R. Part 2002, and it applies to information the government must protect but has not classified — law enforcement data, critical infrastructure details, privacy data. CUI faces far less oversight than classification, which makes it a quieter route for restricting a failed study. A research record marked CUI is not classified and is not subject to ISCAP review, but it may still be exempt from FOIA disclosure under one of the nine exemptions.

Can I force release of a classified failed study?

The realistic routes are administrative and slow. A mandatory declassification review request under E.O. 13526 section 3.5 goes to the owning agency, which must respond; an appeal goes to ISCAP. A FOIA request runs under 5 U.S.C. § 552 with administrative appeal and eventual litigation. The FOIA Improvement Act of 2016 foreseeable harm standard helps against Exemption 5 withholdings by requiring a specific harm, and asking for the classification guide and source document exposes derivative markings that lack a real basis. No route guarantees release, and the 25-year automatic declassification default at section 3.3 does not apply to records not designated historically valuable.

Does foreign involvement in a failed study affect whether it can be classified?

Sometimes, and this is an underappreciated edge case. The foreign relations category at E.O. 13526 section 1.4(d) covers "confidential sources" and foreign activities, so a failed study conducted with a foreign collaborator or funded through a foreign program can be caught by that category even if the scientific content is unremarkable. The same failure with an entirely domestic team may stay unclassified. The practical implication is that international research partnerships need a classification review at the contract stage, not after the results come in, because retroactive classification of an already-published finding creates its own legal exposure under the order's reclassification rules at section 1.7(c).

What happens to a federal employee who discloses a classified failed study?

Disclosure of properly classified information can trigger administrative sanctions, loss of clearance, and in egregious cases criminal exposure under 18 U.S.C. § 798, which covers disclosure of classified cryptographic and communications intelligence information. Whistleblower protections under the Whistleblower Protection Act of 1989 and the Intelligence Community Whistleblower Protection Act of 1998 are narrow and generally require disclosure through designated channels such as an Inspector General rather than public release. A federal scientist who believes a failure has been improperly classified should route a challenge through the agency's classification challenge process at E.O. 13526 section 1.8, which provides a formal channel for authorized holders to challenge a marking without exposing themselves to sanctions.

This article is general educational information about federal classification policy, not legal advice. Classification decisions, FOIA appeals, and whistleblower routes depend on the specific agency, program, security agreement, and document involved; consult a qualified attorney with federal information law experience before disclosing or challenging classified material. If you face an urgent safety or legal emergency, contact local emergency services or counsel immediately.