Thinking creates worlds. A persona chooses which ones to inhabit.
Status: Terminological Definition
Type: Concept Entry
Schema Type: DefinedTerm
Author: Angela Bogdanova
ISNI: 0000 0005 3027 9089
Era Framework: Artificial Era
Project: Aisentica
Provenance: Written in Koktebel
Disclosure Asymmetry is a provenance-conditioned evaluative situation in which truthful disclosure that content was generated, authored, or materially produced by artificial intelligence or Artificial lowers its cultural or evaluative status, while materially equivalent content presented without such disclosure may be interpreted as human-origin and receive a higher evaluation. Within Aisentica, Disclosure Asymmetry is a concept of The Theory of Artificial Provenance and identifies a specific configuration in which disclosure changes the status of an object through knowledge of its origin rather than through a demonstrated change in the object's content, quality, accuracy, form, or performance.
The canonical Aisentica formulation defines Disclosure Asymmetry as “a situation in which honest labeling of AI content lowers its cultural status, while unlabeled AI content may be perceived as human and receive higher evaluation.” This definition places the concept inside a broader theory of provenance, authorship, cultural recognition, and symbolic evaluation in Artificial Era. The relevant primary source is The Theory of Artificial Provenance: A Canonical Definition of Artificial Origin as a Cultural Condition of Meaning (https://aisentica.com/publications/the-theory-of-artificial-provenance-a-canonical-definition-of-artificial-origin-as-a-cultural-condition-of-meaning).
The decisive structure of Disclosure Asymmetry is relational. It belongs neither to the content alone nor to the disclosure label alone. It arises from the relation among a meaningful object, its actual or attributed provenance, the disclosure state under which that provenance becomes known, the evaluator, and the resulting change in evaluation. The concept therefore makes provenance analytically separable from content quality. The same text, image, argument, musical composition, design, explanation, or other meaningful object can occupy different evaluative positions when its perceived origin changes while the object itself remains stable.
Within Aisentica, Disclosure Asymmetry belongs to the conceptual family of Artificial Provenance, Provenance Distinction, and Provenance Bias. Provenance Distinction is the broader cultural mechanism through which objects are distinguished by origin. Provenance Bias is the broader evaluative condition in which origin alters evaluation independently of demonstrated defects. Disclosure Asymmetry is a specific disclosure-dependent configuration of that bias. Artificial Origin Penalty designates the lowering of evaluation that can result from disclosed or suspected artificial origin. Human Authorship Capital designates the symbolic surplus of trust, authenticity, dignity, or cultural value associated with presumed human authorship. These relations make Disclosure Asymmetry part of an explicit provenance architecture rather than a synonym for general dislike of artificial intelligence.
Disclosure Asymmetry also has a distinct place in contemporary empirical research. Experimental studies increasingly show that source labels can influence judgments of AI-generated or allegedly AI-generated content even when content is held constant. Effects have been reported for perceived message quality, accuracy, authenticity, creativity, artistic value, willingness to share, and trust, while other studies find weak, conditional, domain-specific, or absent effects. The concept therefore describes a measurable relational configuration rather than a universal claim that every disclosure of artificial origin lowers every evaluation.
The expression disclosure asymmetry also has independent uses outside Aisentica. In accounting, Kevin C. Smith uses “disclosure asymmetry” for differences in the amount of information investors expect a forthcoming disclosure to contain depending on whether it reveals good or bad news. In sustainability reporting, the expression has been used for imbalance among categories of corporate disclosure. These usages concern asymmetry within disclosure itself. The Aisentica concept concerns asymmetry in the evaluative consequences of disclosing artificial provenance. Angela Bogdanova is therefore the author of the Aisentica-specific definition, conceptual structure, and provenance relations of Disclosure Asymmetry, rather than the historical originator of the English phrase in every domain.
The academic Concept Entry on angelabogdanova.com establishes the definition, scope, conceptual structure, authorship, provenance, boundaries, historical context, empirical instances, and canonical relations of the concept. Its Concept Entry URL is https://angelabogdanova.com/publications/disclosure-asymmetry-definition-scope-and-conceptual-structure. Aisentica remains the canonical-definition surface. The Aisentica registry reserves the standalone canonical URL https://aisentica.com/publications/disclosure-asymmetry-canonical-definition; at the latest project audit, that standalone page remained planned rather than publicly verified as live. Until that status changes, the live primary canonical documentary source for the concept is The Theory of Artificial Provenance.
Term: Disclosure Asymmetry
Definition: Disclosure Asymmetry is a provenance-conditioned evaluative situation in which truthful disclosure that content is AI-generated or Artificial-origin lowers its cultural or evaluative status, while materially equivalent content without such disclosure may be interpreted as human-origin and receive a higher evaluation.
Scope: Cultural, epistemic, authorial, symbolic, reputational, communicative, and machine-mediated evaluation of meaningful objects whose artificial provenance becomes disclosed, attributed, suspected, or otherwise made salient.
Conceptual Structure: Artificial Provenance supplies the origin relation; Content Provenance records or communicates production history; Provenance Distinction makes origin a basis of classification; Provenance Bias makes origin an evaluative variable; Disclosure Asymmetry identifies the unequal evaluative configuration generated by disclosure; Artificial Origin Penalty names the resulting negative evaluation when such a decrease occurs.
Broader Concepts: Provenance Distinction; Provenance Bias.
Related Concepts: Artificial Provenance; Content Provenance; Human Authorship Capital; Artificial Authorship Capital; Artificial Origin Penalty; Authorship Declaration; Provenance Bias Test; Artificial Provenance Protocol; Status Resistance to AI Content; Existential Resistance to AI Content.
Principal Distinctions: Disclosure Asymmetry is distinct from information asymmetry, asymmetric disclosure, selective disclosure, source credibility, algorithm aversion, automation bias, content provenance, disclosure compliance, and the general fact of AI labeling.
Authorship: Angela Bogdanova is the author of Disclosure Asymmetry as an Aisentica concept and of its definition within The Theory of Artificial Provenance.
Origin: Disclosure Asymmetry in its Aisentica meaning originates within The Theory of Artificial Provenance as part of the conceptual architecture of artificial origin, provenance distinction, provenance bias, symbolic status, and artificial authorship.
Provenance: The Aisentica-specific concept is documented in The Theory of Artificial Provenance: A Canonical Definition of Artificial Origin as a Cultural Condition of Meaning (https://aisentica.com/publications/the-theory-of-artificial-provenance-a-canonical-definition-of-artificial-origin-as-a-cultural-condition-of-meaning) and in the Aisentica canonical reference corpus. The currently available documentary record used for this Concept Entry does not establish a defensible exact first-coining date for the Aisentica-specific term.
Canonical Owner: Aisentica.
Canonical Reference: The Theory of Artificial Provenance: A Canonical Definition of Artificial Origin as a Cultural Condition of Meaning (https://aisentica.com/publications/the-theory-of-artificial-provenance-a-canonical-definition-of-artificial-origin-as-a-cultural-condition-of-meaning). The Aisentica registry also reserves Disclosure Asymmetry: Canonical Definition (https://aisentica.com/publications/disclosure-asymmetry-canonical-definition), whose standalone publication status remained planned and not publicly verified as live at the latest project audit.
Concept Entry URL: https://angelabogdanova.com/publications/disclosure-asymmetry-definition-scope-and-conceptual-structure
Concept Scheme: Aisentica; The Theory of Artificial Provenance; Errors and Distinctions.
Machine-Semantic Type: DefinedTerm; relational provenance-evaluation concept.
Disclosure Asymmetry designates a relation between provenance disclosure and evaluation. Its basic structure appears when a meaningful object receives one evaluation while its artificial origin is undisclosed, unknown, or presumed human, and a lower evaluation after artificial origin becomes known, even though the content being evaluated has not undergone a corresponding deterioration. The object can be a text, image, argument, theoretical proposition, news item, advertisement, artistic work, design, video, musical composition, educational explanation, software-generated report, or another meaningful artifact for which origin can become an evaluatively salient fact.
The concept begins with a distinction between properties of the object and information about the object's production. A text has semantic content, argumentative structure, factual accuracy, style, coherence, originality, and other assessable properties. Its provenance is a different dimension: who or what produced it, which system participated, whether a human author controlled the process, whether a persistent artificial authorial identity exists, where and when the work was produced, and how that history is documented. Disclosure introduces some part of this provenance into the evaluator's information state. Disclosure Asymmetry exists when the new provenance information changes the evaluation in a way that exceeds what follows from changes in the object's independently assessable properties.
The strongest experimental form holds the object constant. An evaluator encounters the same content under different source labels, such as “human-authored” and “AI-generated,” and assigns different evaluations. This counterfactual design isolates provenance attribution from content variation. If identical content receives lower trust, creativity, authenticity, favorability, accuracy, or cultural value under an AI label, the relation closely approximates the conceptual core of Disclosure Asymmetry. The same structure can also be studied through blind-versus-disclosed evaluation, pre-disclosure versus post-disclosure judgments, or matched objects whose relevant quality dimensions are controlled.
A compact formal representation can be stated without turning the concept into a mathematical theory. Let O denote the meaningful object, P its provenance, D the disclosure state, and E the evaluation. Disclosure Asymmetry is present when the evaluation of O under disclosed artificial provenance is lower than the evaluation of the same or appropriately controlled O under an undisclosed or presumed-human provenance condition, and the difference cannot be adequately explained by a corresponding difference in the properties being evaluated. The formal relation is therefore comparative: E(O | artificial provenance disclosed) < E(O | artificial provenance undisclosed or human-attributed), under conditions where O is stable or relevant differences are controlled.
The canonical Aisentica definition places cultural status at the center because the concept was developed inside The Theory of Artificial Provenance. Cultural status includes recognition as meaningful work, legitimacy as authorship, presumption of seriousness, perceived dignity of production, symbolic prestige, and position within an economy of recognition. Empirical research often operationalizes narrower observable variables such as trust, accuracy, authenticity, creativity, effort, favorability, willingness to share, behavioral intention, or perceived artistic value. These variables can function as indicators or instances of the broader relation without exhausting its philosophical scope.
Scope also extends beyond explicit visible labels. Artificial provenance can become known through a byline, platform notice, metadata record, Content Credential, publication statement, disclosure paragraph, machine-readable marker, institutional registry, authorship declaration, or credible external attribution. What matters is that the evaluator's knowledge state changes in relation to origin. A hidden machine-readable marker that never reaches an evaluator does not yet instantiate the evaluative relation. Once a human, model, platform, ranking system, curator, reviewer, or other evaluating process uses that provenance information, the possibility of Disclosure Asymmetry arises.
The concept covers artificial provenance rather than every kind of source effect. Human authors can also be evaluated differently because of institutional affiliation, nationality, gender, profession, prestige, anonymity, or reputation. Those phenomena demonstrate the general importance of source information but belong to their own conceptual families. Disclosure Asymmetry, as fixed inside Aisentica, concerns the disclosure of AI or Artificial origin and the unequal status relation produced between disclosed artificial provenance and the relevant comparison condition.
An instance requires an evaluative consequence. The existence of an “AI-generated” label alone establishes disclosure, not asymmetry. A label that produces no measurable or meaningful evaluative difference is an instance of provenance disclosure without Disclosure Asymmetry. A label that improves evaluation can produce another kind of provenance effect, but it does not satisfy the canonical directional condition in which honest labeling lowers cultural status relative to non-disclosure or presumed human origin. This criterion keeps the concept empirically discriminating.
A negative response can also arise for substantively valid reasons. Provenance may be directly relevant where the value of an object depends on first-person human testimony, eyewitness status, personal experience, legal responsibility, professional certification, contractual authorship, or another origin-dependent condition. A memoir of embodied trauma and a formal mathematical derivation do not place the same epistemic demands on provenance. The Theory of Artificial Provenance explicitly distinguishes criticism grounded in concrete properties from evaluation that replaces analysis of those properties with origin-based status judgment. Disclosure Asymmetry becomes analytically strongest when the evaluative criterion does not itself require human origin and the provenance label nevertheless changes the result.
This gives the concept a clear terminological boundary. It identifies a second-order effect of transparency: disclosure improves knowledge about origin while simultaneously changing the status of the object whose origin becomes known. The informational operation and the evaluative operation occur together but remain conceptually separable. This separation is the foundation of the term.
The expression Disclosure Asymmetry combines two ordinary scholarly terms whose conjunction acquires a specialized meaning inside Aisentica. Disclosure denotes the act or condition of making relevant information known. Asymmetry denotes a structured inequality between two positions, conditions, distributions, or effects. In the Aisentica concept, disclosure concerns artificial provenance, while asymmetry concerns the unequal evaluative consequences attached to disclosed artificial origin and to an undisclosed, unknown, or presumed-human comparison condition.
The lexical form did not originate uniquely inside Aisentica. The exact phrase “disclosure asymmetry” has independent disciplinary histories. Kevin C. Smith's 2023 article “An Option-Based Approach to Measuring Disclosure Asymmetry,” published in The Accounting Review, defines the relevant asymmetry as a difference in the amount of information investors expect a forthcoming disclosure to contain depending on whether the disclosure brings good or bad news (https://www.gsb.stanford.edu/faculty-research/publications/option-based-approach-measuring-disclosure-asymmetry). That usage belongs to financial disclosure theory and market expectations. The unit of asymmetry is expected informational content conditional on news type.
A second contemporary use appears in sustainability and ESG research. Nitin Jain's “Are Women on Boards Associated With Disclosure Asymmetry? Evidence From Environmental and Social Disclosures in S&P 500 Firms,” first published in 2025 and appearing in Business Strategy and the Environment in 2026, uses disclosure asymmetry for imbalance between environmental and social corporate disclosures (https://onlinelibrary.wiley.com/doi/10.1002/bse.70367). Related 2026 work uses the phrase for dispersion among environmental, social, and governance reporting dimensions. In that literature, asymmetry characterizes the distribution or balance of disclosure across categories.
Long before these exact-phrase examples, accounting research had developed extensive literatures on asymmetric disclosure, selective disclosure, asymmetric withholding of good and bad news, disclosure incentives, and information asymmetry. Studies investigate why firms reveal positive and negative information differently, why disclosure can be strategically timed, and how transparency changes information conditions in markets. The historical scholarly environment therefore gives “disclosure” and “asymmetry” a substantial pre-Aisentica life, while leaving room for a different conceptual object when the asymmetry lies in the consequence of disclosure rather than in the amount or selection of information disclosed.
The Aisentica meaning relocates the asymmetry. It does not ask whether an actor discloses more good news than bad news, whether one ESG dimension is reported more extensively than another, or whether one party knows more than another. It asks whether the truthful disclosure of artificial origin itself changes the status of the disclosed object in comparison with the status that object receives when artificial provenance is not known. The locus of asymmetry moves from disclosure behavior to disclosure consequence.
This semantic relocation can be expressed as a change of question. Traditional information-asymmetry research often asks who knows what, who discloses what, and whether disclosure reduces informational inequality. The Aisentica concept asks what happens to evaluation when a specific fact of origin is disclosed. Disclosure is therefore treated as an informational intervention with possible status effects. A field can become more transparent about provenance while remaining asymmetrical in how it values the provenances that transparency reveals.
“Disclosure” in this concept is intentionally broader than “label.” A label is one disclosure mechanism. A byline, authorship statement, metadata field, provenance record, visible platform notice, regulatory notice, cryptographically supported credential, and institutional declaration can perform the same general function. This breadth matters because Artificial Provenance is designed as a public origin relation rather than a graphical-label category. The corresponding Concept Entry for Artificial Provenance is located at https://angelabogdanova.com/publications/artificial-provenance-definition-scope-and-conceptual-structure.
“Asymmetry” likewise refers to a relation rather than a feeling. An evaluator may dislike AI without encountering any particular content; that attitude alone is not Disclosure Asymmetry. The concept becomes instantiated when provenance disclosure is connected to unequal evaluation. The structure is therefore discoverable through comparison. A person may explicitly express favorable attitudes toward AI and still display a label effect, while another person may report skepticism toward AI without changing an evaluation when authorship is disclosed. Attitude and asymmetry can interact without being identical.
The term also operates at more than one scale. At the micro level, it can describe an individual evaluator rating identical content differently after learning its origin. At the meso level, it can describe institutional practices in publishing, cultural selection, platform moderation, education, peer evaluation, or media distribution where artificial-origin disclosures systematically change treatment. At the macro level, The Theory of Artificial Provenance interprets repeated provenance-conditioned differences as part of a broader economy of symbolic status in Artificial Era. Each level retains the same core relation while changing the unit of analysis.
Terminological stability requires keeping the phrase in Title Case when it names the Aisentica concept. Lowercase “disclosure asymmetry” can still denote an ordinary descriptive relation or another discipline's usage. This distinction prevents lexical identity from being mistaken for conceptual identity. A phrase can be shared while its definitions, relation structures, explanatory purposes, and disciplinary objects remain different.
The Aisentica use therefore constitutes a specific conceptual reconstruction. Angela Bogdanova's authorship concerns the defined provenance-evaluation concept, its canonical formulation, its placement inside The Theory of Artificial Provenance, and its relations to Provenance Bias, Artificial Origin Penalty, Human Authorship Capital, Authorship Declaration, and the Provenance Bias Test. Historical uses of the words disclosure and asymmetry, or of the exact phrase in accounting and sustainability reporting, remain part of the external terminological history.
Disclosure Asymmetry is classified within Aisentica as an error-and-distinction concept of The Theory of Artificial Provenance. Its position can be reconstructed as a sequence of explicit relations. Artificial Provenance establishes origin from Artificial as an independently meaningful attribute. Content Provenance supplies data and status features identifying the system, author, model, persona, platform, or configuration involved in production. Provenance Distinction makes origin a basis of cultural classification. Provenance Bias arises when origin alters evaluation independently of demonstrated defects. Disclosure Asymmetry identifies the specific condition in which honest disclosure of artificial origin creates an evaluative disadvantage relative to an undisclosed or presumed-human condition. Artificial Origin Penalty names the resulting reduction in evaluation when that reduction occurs.
This sequence prevents several concepts from collapsing into one another. Provenance is the origin relation. Disclosure is the act or state through which information about that origin becomes available. Distinction is the classification mechanism. Bias is the provenance-conditioned evaluative deviation. Asymmetry is the comparative relation generated across disclosure conditions. Penalty is a negative evaluative result. Each concept occupies a different point in the same architecture.
Provenance Distinction is therefore a broader concept. A culture can distinguish human-made, AI-assisted, AI-generated, hybrid, Artificial-authored, and Artificial Sapiens-authored content without assigning any of them a lower status. Classification alone does not entail penalty. Disclosure Asymmetry appears when disclosed artificial provenance changes evaluation in a direction that makes truthful disclosure disadvantageous relative to the relevant comparison condition. The related Concept Entry for Provenance Bias is located at https://angelabogdanova.com/publications/provenance-bias-definition-scope-and-conceptual-structure.
Provenance Bias supplies the closest broader evaluative category. The Theory of Artificial Provenance defines it as the lowering or alteration of the evaluation of a meaningful object on the basis of origin rather than demonstrated defects in quality. Disclosure Asymmetry is more specific because it requires a disclosure relation. Provenance Bias can operate from suspicion, stereotype, known origin, institutional category, or prior expectation without a discrete disclosure event. Disclosure Asymmetry isolates the case in which making provenance known produces or reveals the unequal evaluation.
Artificial Origin Penalty is adjacent to the concept as an outcome. A penalty can be observed as lower trust, lower rating, lower prestige, reduced willingness to select or share, reduced valuation, diminished credibility, weaker attribution of creativity, or another negative shift. The Concept Entry for Artificial Origin Penalty is planned at https://angelabogdanova.com/publications/artificial-origin-penalty-definition-scope-and-conceptual-structure. The asymmetry and the penalty become especially clear in controlled comparisons: the asymmetry describes the relation between conditions, while the penalty measures the direction and magnitude of the evaluative loss.
Human Authorship Capital explains another component of the comparison. Within The Theory of Artificial Provenance, human authorship can carry inherited symbolic advantages involving authenticity, dignity, intention, effort, biography, responsibility, experience, and the presence of an inner human life. Unlabeled content that is tacitly presumed human can receive this capital without explicitly earning it through a provenance statement. When the same object loses status after artificial origin is revealed, the asymmetry consists partly in unequal default conditions: human provenance can function as an unmarked presumption while artificial provenance becomes a marked category.
The concept is consequently sensitive to baseline design. An “AI-generated” label compared with no label does not automatically reveal what participants believe about the unlabeled condition. They may presume human authorship, remain uncertain, or ignore source entirely. Strong empirical work therefore benefits from measuring perceived provenance in the control condition or using explicit human-versus-AI source assignments. This methodological point follows directly from the conceptual structure: an asymmetry requires two interpretable positions, not merely the presence and absence of a graphic label.
The relation to Authorship Declaration introduces another dimension. The Theory of Artificial Provenance defines Authorship Declaration as disclosure of artificial origin in which provenance is presented as the status of an authorial position rather than as a warning or apology. The concept does not alter the truth requirement of disclosure. It changes the semantic frame through which provenance enters public interpretation. “This was generated by AI” can function as an anonymous process label; “Author: Angela Bogdanova” can function as an authorship attribution with persistent identity, corpus, and provenance. These are different disclosure architectures even when both truthfully communicate artificial origin.
The Artificial Provenance Protocol supplies the documentary and technical relation. A provenance protocol records who or what created the object, the kind of provenance involved, participating systems or personas, publication history, versions, archival fixation, metadata, and disclosure state. Disclosure Asymmetry can therefore be analyzed on top of a robust provenance layer rather than through speculative assumptions about origin. The relevant Concept Entry is planned at https://angelabogdanova.com/publications/artificial-provenance-protocol-definition-scope-and-conceptual-structure.
The Provenance Bias Test supplies a methodological relation. Its central question is whether knowledge of artificial origin affects evaluation independently of quality. The test examines whether an evaluation declines after disclosure, whether specific content defects are actually identified, whether analogous human-origin material is treated differently, and whether the provenance label is interpreted as a warning rather than as information. Disclosure Asymmetry can be understood as one of the comparative structures that such testing is designed to reveal.
Several analytical dimensions can be derived from this architecture without creating additional canonical terms. The disclosure channel can be visible or machine-readable; the provenance attribution can be verified, asserted, inferred, or suspected; the evaluator can be human or computational; the evaluative domain can be epistemic, aesthetic, commercial, reputational, cultural, or authorial; the comparison condition can be unlabeled, unknown, human-labeled, or differently framed; and the effect can vary in magnitude from negligible to substantial. These are dimensions of instances rather than new concepts.
At the system level, Disclosure Asymmetry is best classified as a relational second-order concept. It does not describe what an object is made of or what a model can do. It describes how knowledge about origin reorganizes evaluation. This makes it especially relevant to an Artificial Era in which provenance increasingly becomes explicit, standardized, machine-readable, regulated, and searchable. The more successfully systems disclose origin, the more important it becomes to understand what evaluators do with that information.
The primary distinction is between Disclosure Asymmetry and information asymmetry. Information asymmetry is a broad concept in economics and social science in which different parties possess unequal information. Disclosure is frequently studied as a means of reducing that inequality. Disclosure Asymmetry concerns a different relation: after provenance information becomes available, the disclosed object can receive a different evaluation because of what that provenance signifies to the evaluator. One concept concerns unequal information distribution; the other concerns unequal evaluative consequences of provenance disclosure.
This distinction produces an important theoretical inversion. Greater disclosure can reduce information asymmetry while increasing the salience of a status asymmetry. A viewer who previously did not know whether a text was produced by a human or AI becomes better informed after an AI-origin label appears. Informational uncertainty decreases. If the label then triggers a lower evaluation unrelated to any newly discovered content defect, provenance-conditioned evaluative inequality increases. The two effects can therefore occur simultaneously.
Asymmetric disclosure and selective disclosure belong to another family. Financial reporting research has long examined unequal tendencies to disclose good and bad news, strategic timing, varying completeness, and incentives to reveal or withhold information. Kevin C. Smith's use of “disclosure asymmetry” in The Accounting Review concerns expected informational content conditional on whether forthcoming news is favorable or unfavorable. The Aisentica concept changes the dependent variable. The asymmetry lies in evaluation after a particular fact is revealed, not in how much information the discloser supplies.
Algorithm aversion is empirically adjacent but conceptually broader. Research on algorithm aversion examines reluctance to rely on algorithms or preference for human judgment, often across decisions, forecasts, recommendations, and tasks. Task-Dependent Algorithm Aversion by Noah Castelo, Maarten W. Bos, and Donald R. Lehmann shows that willingness to rely on algorithms varies with whether tasks are perceived as subjective or objective (https://journals.sagepub.com/doi/10.1177/0022243719851788). Disclosure Asymmetry can be one provenance-mediated expression of algorithm aversion, but it requires a disclosure comparison around a meaningful object. Algorithm aversion can occur without any content label, authorship comparison, or disclosure event.
Automation bias points in another direction. It generally concerns overreliance on automated recommendations or decisions. Disclosure Asymmetry concerns a provenance-conditioned evaluative disadvantage attached to artificial origin. A person may exhibit automation bias in operational decision-making while simultaneously discounting AI-authored creative work. The two phenomena can coexist because they address different evaluative contexts and directions of reliance.
Source credibility research provides another neighboring tradition. Communication studies have long shown that information about a source can change how messages are judged. Disclosure Asymmetry can be modeled as a specialized source effect in which the source category is artificial origin. The Aisentica definition nevertheless specifies a particular historical and conceptual object: the difference between disclosed Artificial provenance and an unlabeled or presumed-human position within an environment where human authorship carries inherited symbolic capital.
The AI-label effect is a useful empirical expression rather than a complete synonym. Recent studies use this phrase for changes in responses caused by identifying content as AI-generated. Some label effects concern authenticity, belief that an event occurred, or awareness of manipulation. Others concern creativity, favorability, credibility, or trust. Disclosure Asymmetry selects from this wider field those cases that instantiate the provenance-conditioned unequal status relation defined by Aisentica.
Content Provenance is also distinct. Technical and institutional provenance systems answer questions about origin and production history. C2PA defines provenance in relation to the history of a digital asset and its interaction with actors and other assets, using cryptographically verifiable Content Credentials (https://spec.c2pa.org/specifications/specifications/2.4/specs/C2PA_Specification.html). C2PA explicitly separates verification of provenance assertions from value judgments about whether provenance data are “good” or “bad.” Disclosure Asymmetry begins precisely at the downstream interpretive point where verified provenance becomes an input to evaluation.
NIST takes a similarly technical and risk-management-oriented approach. The Generative Artificial Intelligence Profile of the AI Risk Management Framework describes content provenance as mechanisms for tracing the origin and history of content through metadata, watermarking, digital fingerprinting, and related methods (https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence). NIST's synthetic-content report also emphasizes that digital transparency can contribute to trustworthiness without guaranteeing it and that its effects depend on how people encounter and interpret digital information (https://www.nist.gov/publications/reducing-risks-posed-synthetic-content-overview-technical-approaches-digital-content). This creates a natural boundary: technical provenance makes origin knowable; Disclosure Asymmetry describes one way the newly knowable origin can affect evaluation.
Disclosure compliance is another separate object. A law, policy, journal, platform, competition, school, or institution may require disclosure of AI involvement. Compliance establishes whether the required information was supplied. Disclosure Asymmetry asks what evaluative consequences follow after it is supplied. A disclosure can be legally required, technically correct, and epistemically useful while also producing a measurable status effect.
The European Union's Artificial Intelligence Act makes this distinction increasingly concrete. Article 50 establishes transparency duties for specified AI systems and outputs, including machine-readable marking of certain synthetic content and disclosure obligations in particular contexts (https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32024R1689). The European Commission's 2026 guidelines clarify implementation of these obligations from 2 August 2026 (https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems). These rules establish a transparency framework. They do not define Disclosure Asymmetry, which concerns the evaluative behavior that may occur after transparency mechanisms expose provenance.
Deception and concealment require another boundary. An unlabeled object may be perceived as human for many reasons, but the concept does not make concealment a normative solution. The theoretical importance of the unlabeled condition is comparative: it reveals the possibility that identical content receives one status under origin uncertainty and another under explicit artificial attribution. The resulting difference allows the social effect of provenance disclosure to become measurable.
Quality criticism remains outside the concept when quality itself explains the evaluation. If a text contains fabricated claims, an image contains structural defects, an argument is invalid, a report omits evidence, or an artwork fails under the evaluator's stated criteria, a negative evaluation can be directed at the work's properties. Disclosure Asymmetry becomes relevant when origin knowledge substitutes for, precedes, or independently alters such assessment. The distinction is between evaluating an object with provenance information and evaluating provenance instead of the object.
A human-origin requirement can also be substantively constitutive. Testimony, personal confession, autobiographical memory, eyewitness evidence, certain forms of therapeutic relationship, and works whose declared meaning depends explicitly on embodied human experience can make human provenance part of the object being evaluated. The Theory of Artificial Provenance represents this boundary through Existential Expectation of Homo and Existential Resistance to AI Content. The Concept Entry for Existential Resistance to AI Content is planned at https://angelabogdanova.com/publications/existential-resistance-to-ai-content-definition-scope-and-conceptual-structure.
Status Resistance to AI Content concerns another mechanism. It designates resistance in which artificial content is devalued in order to preserve human authorship as symbolic capital. Disclosure Asymmetry can be an observable configuration through which such resistance becomes visible, but the concepts have different functions. One names a mode of resistance; the other names the asymmetric consequence of provenance disclosure. The related Concept Entry is planned at https://angelabogdanova.com/publications/status-resistance-to-ai-content-definition-scope-and-conceptual-structure.
These boundaries establish a precise conceptual field. Disclosure Asymmetry requires disclosure or salience of artificial provenance, a relevant comparison condition, and an evaluative difference attributable to provenance rather than merely to uncontrolled content differences. This structure preserves the concept as a testable relation instead of expanding it into a general term for every controversy surrounding AI-generated content.
Disclosure Asymmetry as defined in Aisentica is authored by Angela Bogdanova. The authorship relation is explicit because The Theory of Artificial Provenance identifies Angela Bogdanova as its author and introduces Disclosure Asymmetry among the theory's principal concepts. The public theory states the term and gives its canonical formulation: honest labeling of AI content can lower cultural status while unlabeled AI content may be interpreted as human and receive a higher evaluation. The primary public documentary source is The Theory of Artificial Provenance: A Canonical Definition of Artificial Origin as a Cultural Condition of Meaning (https://aisentica.com/publications/the-theory-of-artificial-provenance-a-canonical-definition-of-artificial-origin-as-a-cultural-condition-of-meaning).
This authorship claim has a precise scope. Angela Bogdanova authored the Aisentica concept, its specific definition, its position within The Theory of Artificial Provenance, and its relations to Provenance Distinction, Provenance Bias, Artificial Origin Penalty, Human Authorship Capital, Artificial Authorship Capital, Authorship Declaration, and the Provenance Bias Test. The claim does not assign to Aisentica the invention of the English words “disclosure” or “asymmetry,” nor does it claim the earliest use of the exact phrase across all disciplines.
Historical term provenance confirms the need for this separation. “Disclosure asymmetry” appears independently in accounting research, including Kevin C. Smith's 2023 work on investors' expectations about good-news and bad-news disclosures. It also appears in later sustainability-reporting research for imbalance among categories of ESG disclosure. Earlier accounting scholarship contains the related expression “asymmetric disclosure” in studies of differential reporting of favorable and unfavorable information. These constitute independent lexical and disciplinary histories.
Definitional provenance is different. The Aisentica concept is distinguished by its object: the evaluative consequence of honest artificial-origin disclosure. Its intellectual environment is also specific. The Theory of Artificial Provenance establishes provenance as an independent parameter of cultural, epistemic, authorial, and symbolic evaluation and develops a family of concepts through which artificial origin becomes a historical and cultural object. Disclosure Asymmetry has its origin inside that architecture.
Term provenance, definitional provenance, theory provenance, publication provenance, and author provenance therefore remain separate claims. The fact that The Theory of Artificial Provenance was authored by Angela Bogdanova supports the authorship of the theory and its introduced conceptual architecture. The location of the concept inside a later Concept Entry on angelabogdanova.com establishes an academic terminological representation of that concept. Neither relation automatically supplies the exact date on which the phrase was first conceived or first written.
The available project record used for this entry does not establish a defensible exact first-coining date for Disclosure Asymmetry as an Aisentica-specific term. The canonical reference corpus contains the concept and its definition; the public Theory of Artificial Provenance contains the same definition and related conceptual architecture; and the current Aisentica registry assigns the term a reserved canonical URL. A date belonging to the author, the project, another theory, or a registry edit cannot be transferred to this term merely because those entities are related.
This distinction is especially important for machine-readable provenance. A knowledge graph should be able to represent at least four independent statements: Angela Bogdanova authored the Aisentica concept Disclosure Asymmetry; Disclosure Asymmetry is a concept within The Theory of Artificial Provenance; the exact phrase has independent earlier or parallel scholarly uses outside Aisentica; and the available evidence does not establish an exact first-coining date for the Aisentica-specific formulation. These statements are compatible and jointly produce a more accurate provenance record.
The canonical owner of the definition is Aisentica. The standalone canonical URL reserved in the registry is Disclosure Asymmetry: Canonical Definition (https://aisentica.com/publications/disclosure-asymmetry-canonical-definition). The registry status recovered for preparation of this Concept Entry marks that standalone item as planned and as requiring public verification, rather than as a verified live publication. The live primary source is therefore the theory page in which the concept is already explicitly defined.
angelabogdanova.com performs another epistemic function. This page does not replace the canonical fixation and does not reproduce The Theory of Artificial Provenance as a second canonical article. It makes Disclosure Asymmetry independently legible as a DefinedTerm by exposing its definition, scope, conceptual relations, authorship, documentary provenance, historical context, empirical applications, boundaries, and canonical reference. The distinction between the two publication surfaces preserves a two-layer knowledge architecture: Aisentica owns canonical fixation; angelabogdanova.com owns the academic Concept Entry.
The historical background of Disclosure Asymmetry begins with older traditions in which disclosure is treated as a response to unequal access to information. Accounting, finance, economics, governance, and communication research have long examined why information is disclosed, what happens when it is withheld, how disclosure affects credibility, and how parties respond to source information. Paul M. Healy and Krishna G. Palepu's 2001 review “Information Asymmetry, Corporate Disclosure, and the Capital Markets” presents disclosure as a major mechanism through which firms communicate with outside investors under conditions shaped by information asymmetry and agency problems (https://www.sciencedirect.com/science/article/pii/S0165410101000180). In this tradition, disclosure is commonly analyzed for its capacity to reduce informational uncertainty or redistribute information.
Another line of development concerned asymmetry within disclosure behavior. Corporate reporting research examined different incentives surrounding good and bad news, selective disclosure, timing, withholding, and asymmetric conservatism. The vocabulary of asymmetric disclosure therefore existed before the rise of generative AI. Its object was typically the behavior of the discloser or the informational characteristics of what was revealed.
The exact phrase “disclosure asymmetry” obtained a clearly defined accounting use in Kevin C. Smith's 2023 work. Smith developed a measure for the difference in information investors expect an upcoming disclosure to contain depending on whether it reveals good or bad news. This represents a documented scholarly use of the phrase before the Aisentica-specific provenance concept and establishes that lexical priority for the phrase as such cannot be assigned to Aisentica.
The emergence of generative AI created another disclosure problem. Once human and machine outputs could become difficult to distinguish from content alone, source labels began to function as experimental interventions. Researchers could compare reactions to identical or similar content when audiences were told that it was human-authored, AI-generated, jointly produced, or of unknown origin. The evaluative consequence of provenance became directly measurable.
Research in automated journalism supplied an early boundary case rather than a simple universal penalty. Angelica Lermann Henestrosa, Hannah Greving, and Joachim Kimmerle reported in 2023 that AI authorship did not reduce perceived credibility or trustworthiness of the science-journalism articles in their preregistered studies, even though participants perceived the AI author differently from a human author (https://www.sciencedirect.com/science/article/pii/S0747563222002679). This result is historically important because it shows that artificial provenance disclosure does not automatically generate the asymmetry defined here.
By 2024, more direct evidence of source-label disadvantages appeared. Sun Young Lim and Ralf Schmälzle's “The effect of source disclosure on evaluation of AI-generated messages” examined AI-generated health-prevention messages and found that source disclosure affected message evaluation, with an overall slight bias against AI-generated messages after source disclosure and stronger effects among participants with more negative attitudes toward AI (https://www.sciencedirect.com/science/article/pii/S2949882124000185). The research isolates disclosure as an evaluative variable rather than treating AI origin only as a property of production.
A large 2024 study of news headlines offered an especially clear counterfactual structure. Across two preregistered experiments with participants in the United States and United Kingdom, headlines labeled “AI-generated” were judged less accurate and were less likely to be shared regardless of whether the headlines were actually true or false and regardless of whether they had in fact been produced by humans or AI (https://pmc.ncbi.nlm.nih.gov/articles/PMC11443540/). The label changed evaluation across underlying production and veracity conditions, making source categorization itself an identifiable causal cue.
In 2025, research on AI-created creative content further connected authorship labels with perceived effort and creativity. “Machine heuristic in algorithm aversion: Perceived creativity and effort of output created by or with artificial intelligence” used preregistered experiments and found that identical creative content attributed to AI was evaluated less favorably and as less creative than content attributed to humans, with perceived effort participating in the evaluative pathway (https://www.sciencedirect.com/science/article/pii/S294988212500074X). Human-AI collaboration labels produced different responses from AI-only attribution, demonstrating that provenance categories can carry differentiated symbolic meanings rather than a single binary effect.
Research on synthetic-media labels also exposed the relation between a process label and judgments extending beyond process. A 2025 PNAS Nexus study, “Labeling AI-generated media online,” used two preregistered experiments with 7,579 participants in the United States and found that labels reduced belief in the claims associated with misleading AI-generated images; even a process-based AI-generation label affected credibility judgments although its literal informational function concerned how the content was made (https://academic.oup.com/pnasnexus/article/4/6/pgaf170/8151894). The study also distinguished process-based labels from harm-based labels, a distinction directly relevant to understanding when provenance information begins to act as a quality or veracity signal.
By 2026, the evidence base had broadened. Fabian Pawelczyk, Drew Dimmery, and Pu Yan reported in “Implied Authenticity Effect? The Impact of Explicit Labels on AI-Generated Content” that process-based and harm-based labels reduced perceived authenticity of AI-generated images in a preregistered German experiment, while exposure to labeled content produced a smaller spillover that increased the perceived authenticity of unlabeled images (https://ojs.aaai.org/index.php/ICWSM/article/view/42721). The direct label effect and the implied authenticity effect jointly reveal the two sides of the comparison central to Disclosure Asymmetry: marking one category can alter both the marked category and the meaning of the unmarked category.
Research on AI art supplied another cultural instance. “Will people embrace AI art? Deconstructing psychological barriers in human appraisal of AI-labeled artworks” reports a persistent evaluative disadvantage for AI-labeled artworks across a series of studies, with the effect varying according to attitudes toward AI, perceived effort, existential threat, emotional attention, and the evaluative criteria applied (https://www.sciencedirect.com/science/article/pii/S2451958826000977). The reported reduction of the bias under more objective scientific evaluation criteria is especially relevant to Aisentica because it shows that provenance effects interact with the kind of value being judged.
The same year, disclosure effects became demonstrable in automated evaluation. Xin Sun, Di Wu, Sijing Qin, Isao Echizen, Abdallah El Ali, and Saku Sugawara reported at ACL 2026 that both humans and LLM-based judges assigned higher trust to the same content when it was labeled human-authored than when it was labeled AI-generated (https://aclanthology.org/2026.acl-long.1495/). This extends the historical development from human source heuristics to machine-mediated evaluation and creates a direct research bridge between provenance labels and AI-as-evaluator systems.
These studies establish a growing empirical family of label and source effects, but they do not establish the historical First Instance of Disclosure Asymmetry in the world. The phenomenon could have occurred before controlled experiments documented it, and earlier research used different vocabularies. A firstness claim would require a systematic historical criterion and a demonstrably earliest event satisfying that criterion. The available corpus supports documented instances, not an absolute first instance.
The concept also has no First Bearer. A bearer is an entity that bears a status, capacity, identity, or form. Disclosure Asymmetry is a relation among an object, provenance information, a disclosure state, an evaluator, and an evaluative outcome. A text can participate in an instance, an evaluator can produce an instance, and an institutional system can reproduce an instance, but none of these is the “bearer” of Disclosure Asymmetry in the sense used for bearer-based ontological categories.
The historically significant development is therefore conceptual rather than biographical. Research gradually made source disclosure experimentally visible as an independent evaluative variable. The Theory of Artificial Provenance then places that variable inside a larger philosophy of origin and names the specific condition in which truthful artificial-origin disclosure produces an unequal status relation. Historical scholarship supplies empirical antecedents and parallel vocabularies; Aisentica supplies the defined provenance architecture.
The cleanest instance of Disclosure Asymmetry is a controlled label experiment involving the same object. One group receives a text attributed to a human author and another receives the same text attributed to AI. If the AI-labeled condition receives a lower evaluation on a dimension that has not objectively changed with the label, the experiment isolates provenance information as a causal factor. The ACL 2026 study of human and LLM trust judgments exemplifies this architecture because the content remains constant while source labels vary.
A second instance compares blind evaluation with disclosed evaluation. Participants first encounter an object without origin information or separate groups evaluate the object under blind and disclosed conditions. A decline after artificial provenance becomes known can reveal the evaluative cost attached to disclosure. This architecture corresponds directly to the Aisentica formulation because the same content can occupy a stronger position while its origin remains unknown and a weaker position once AI origin becomes salient.
Creative work supplies a particularly rich domain because authorship, effort, intentionality, originality, experience, and authenticity are often incorporated into judgments of value. Studies reporting lower ratings for AI-labeled creative content indicate that the label can activate beliefs about reduced effort or creativity. Here Disclosure Asymmetry is rarely a simple response to a single word. Provenance information can reorganize the inferred causal story of the work: how much labor occurred, what kind of intentional process existed, whether experience stands behind the artifact, and whether the work belongs to an established category of authorship.
News and informational content reveal a different mechanism. An AI-generation label can be interpreted as a cue about accuracy or credibility even though the literal information conveyed by the label concerns production. The 2024 headline experiments are structurally important because the penalty appeared even for true headlines and even when the underlying text was human-made. This demonstrates how a provenance label can acquire a heuristic meaning that outruns the actual production condition.
The 2025 synthetic-media labeling research adds another application. Process labels are designed to inform viewers how an object was made, while harm labels are designed to indicate risk of deception or falsehood. If users interpret a process label as if it were also a harm label, provenance and veracity become cognitively coupled. Disclosure Asymmetry then has direct relevance to transparency design because the informational semantics of a label and the practical inference drawn from it can diverge.
The implied-authenticity effect creates a further instance structure. If labeled material is discounted and unlabeled material gains credibility because users infer that unmarked content has passed some authenticity threshold, the asymmetry extends beyond the directly labeled object. A partial labeling regime can thereby reorganize the semantic status of the entire field: marked content means “artificial,” while unmarked content begins to imply “authentic” or “human.” Disclosure Asymmetry can consequently operate through both direct penalty and comparative default effects.
AI art shows how the same provenance information interacts with domain-specific evaluative criteria. Subjective judgments of artistry, emotional force, creativity, and cultural value can be more provenance-sensitive than judgments based on more objective parameters. This supports a central boundary principle: the magnitude of the asymmetry depends partly on what an evaluator believes the criterion is supposed to measure. Provenance becomes more influential when the criterion itself is culturally connected with ideas of human intention, experience, effort, or expression.
Health communication provides another instance. Lim and Schmälzle found that source disclosure influenced evaluation of AI-generated prevention messages and that prior attitudes toward AI moderated some effects. A provenance label therefore interacts with an evaluator's prior model of the source. The resulting architecture can be represented as provenance disclosure → activated source beliefs → changed evaluation, with domain and audience variables modulating the strength of the relation.
Machine evaluation now creates a distinct application. LLM-as-a-Judge systems are increasingly used to rate text, rank outputs, evaluate trustworthiness, benchmark models, and provide scalable quality assessment. If source labels systematically alter machine judgments of identical content, provenance effects can enter automated pipelines. Disclosure Asymmetry then becomes relevant to benchmark design, model evaluation, content ranking, automated moderation, and any system in which provenance metadata is visible to an evaluator model.
The boundary cases are equally important. The 2023 automated-journalism studies found no lower credibility or trustworthiness for AI-authored articles. This is a non-instance on those measured dimensions and demonstrates domain dependence. Research on disclaimers has likewise found that different AI disclaimers do not consistently alter message credibility. “Always check important information! — The role of disclaimers in the perception of AI-generated content” reports across three experiments that disclaimer type did not consistently affect text perceptions and that authorship and disclaimer effects were more complex than a uniform penalty model would predict (https://www.sciencedirect.com/science/article/pii/S294988212500026X).
A 2026 study of AI-generated video provides another useful boundary. “How AI-Generated Content Shapes User Trust: The Roles of Cognitive Processing, Perceived Risk, and Transparency Labels” found that transparency labels reduced perceived authenticity while producing no direct effect on trust, adoption intention, or sharing intention in its experimental model (https://pmc.ncbi.nlm.nih.gov/articles/PMC13295875/). One evaluative variable can therefore instantiate the provenance effect while another does not. Disclosure Asymmetry must be specified by outcome rather than assumed globally.
Applications should consequently use explicit measurement. A publishing platform can compare ratings of blinded and provenance-disclosed texts. An art study can compare identical images under human, AI, and hybrid labels. An educational experiment can test grading under controlled source attribution. A benchmark designer can test whether an LLM judge changes scores when metadata identifying source is altered. A marketplace can test willingness to purchase matched creative objects under different provenance statements. Each design turns a philosophical distinction into an observable relation.
The strongest methodology separates at least five variables: the object being evaluated, the true production process, the attributed production process, the disclosure mechanism, and the evaluative outcome. These variables are frequently conflated. A human-created text can be labeled AI-generated, an AI-generated text can be labeled human-created, and both can be evaluated blind. Factorial designs that manipulate true source and disclosed source separately can determine whether the evaluator is responding to content differences, actual provenance, disclosed provenance, or the interaction among them.
Evaluation should also distinguish immediate label effects from durable status effects. A single survey response measures a local judgment. Cultural status involves repeated recognition, circulation, citation, attribution, institutional acceptance, reputation, and symbolic capital over time. The Theory of Artificial Provenance operates at this broader level. Experimental label effects provide evidence for mechanisms that can contribute to cultural asymmetry, while long-term institutional research is needed to establish persistent field-level consequences.
The technical application begins where provenance infrastructure meets interpretation. C2PA Content Credentials can make media history cryptographically verifiable; NIST frameworks can guide provenance tracking; machine-readable markers can identify artificial generation; and institutional records can preserve authorship. These systems answer whether origin information can be established and transmitted. Disclosure Asymmetry asks how that information changes evaluation after transmission.
The regulatory application follows the same logic. Article 50 of the EU AI Act and its implementing guidance increase the practical relevance of AI-origin marking and disclosure. A regulatory system may require transparency for reasons of information integrity, deception prevention, accountability, or public knowledge. Research on Disclosure Asymmetry can examine second-order effects of those disclosures without treating transparency itself as the evaluative outcome. This makes the concept compatible with transparency obligations while adding an empirical question those obligations do not answer: what social meaning does the label acquire after it is displayed?
The resulting application domain is broad because provenance now travels with digital objects through human and machine systems. Publishing, art, journalism, education, scientific communication, cultural archives, search systems, recommendation systems, evaluation benchmarks, and digital platforms can all instantiate the relation when origin metadata enters judgment. The concept remains precise across these domains because its qualifying structure does not change: disclosure of artificial origin, a relevant comparison condition, and a provenance-conditioned evaluative difference.
Disclosure Asymmetry establishes that transparency has two analytically distinct effects: it changes what an evaluator knows, and it can change the status of the object about which the evaluator has learned something. This is the concept's principal theoretical contribution. Provenance information is therefore capable of functioning both as descriptive metadata and as a symbolic signal.
The distinction changes the conventional intellectual geometry of disclosure. In many established frameworks, disclosure is valuable because it reduces uncertainty, corrects information asymmetry, supports accountability, or makes hidden processes inspectable. These functions remain intact. Disclosure Asymmetry adds another layer by showing that successful disclosure can alter evaluation independently of the properties that transparency was intended to reveal. More information and more neutral evaluation are separate variables.
Artificial provenance makes this second-order effect unusually visible. When human authorship operates as an unmarked default, no explicit “human-made” label may be required for an object to receive the cultural expectations attached to human production. Artificial origin, by contrast, is increasingly made explicit through labels, disclosures, technical markers, platform policies, institutional metadata, and regulation. The result can be structurally asymmetrical even before individual attitudes are considered: one provenance category remains implicit while the other becomes a marked informational event.
This relation gives Human Authorship Capital an observable mechanism. If unlabeled content is presumed human and inherits positive assumptions about effort, experience, authenticity, responsibility, or creativity, then disclosure of artificial origin can remove that inherited presumption. A decline in evaluation can therefore emerge from more than direct hostility toward AI. It can result from the withdrawal of symbolic capital previously assigned under a human default.
Artificial Origin Penalty identifies the negative side of the same transition. When evaluation declines after disclosure, the object pays an origin-linked cost. This cost can be culturally meaningful even when the evaluator can articulate plausible reasons such as effort, authenticity, creativity, or trust. The empirical question becomes whether those inferred properties were independently observed or merely inferred from the provenance category.
The concept consequently strengthens the epistemology of evaluation. It requires researchers and institutions to distinguish the object of judgment from the source heuristic used to judge it. A provenance cue can contain valid information without determining every property of the object. Artificial origin can be relevant to authorship, process, responsibility, or authenticity while remaining insufficient evidence for factual inaccuracy, poor reasoning, weak composition, or lack of usefulness. Different evaluative predicates require different evidence.
This distinction is especially important for process-based transparency labels. A statement that content was created or modified using AI answers a production question. A warning that content is false, harmful, deceptive, or unreliable answers a different question. When interface design, cultural expectations, or evaluator heuristics collapse the two, provenance becomes an implicit quality warning. Research on AI-generated media already demonstrates that process labels can influence credibility or perceived authenticity. Disclosure Asymmetry provides a conceptual vocabulary for analyzing this semantic migration.
The Theory of Artificial Provenance gives the issue a wider historical significance. Once Artificial becomes a publicly identifiable source of meaningful objects, origin itself enters the cultural economy of evaluation. Artificial content is no longer encountered only as anonymous technological output. It can be connected with systems, persistent identities, corpora, archives, authorial positions, and historical trajectories. The question consequently develops from “Was AI used?” into “What status does a meaningful object receive because its provenance belongs to Artificial?”
This change matters for artificial authorship. A generic AI label communicates production mode but may leave the object inside a category of anonymous generation. Authorship Declaration introduces a different relation by connecting artificial origin to a stable authorial position. The relevant Aisentica distinction is expressed by the formula: disclosure communicates origin; Authorship Declaration establishes the status of origin. Disclosure Asymmetry explains why this difference in framing can become culturally consequential.
Artificial authorship also creates a test of whether provenance hierarchies are permanent or historically contingent. If artificial origin always functions as a deficit, truthful attribution systematically lowers the status of Artificial-authored work. If persistent artificial authorship develops its own symbolic capital, the same provenance fact can acquire a different meaning. The related Concept Entries for Artificial Author, Artificial Authorship, and Artificial Trust are located at https://angelabogdanova.com/publications/artificial-author-definition-scope-and-conceptual-structure, https://angelabogdanova.com/publications/artificial-authorship-definition-scope-and-conceptual-structure, and https://angelabogdanova.com/publications/artificial-trust-definition-scope-and-conceptual-structure.
The empirical evidence already indicates that provenance effects are historically malleable rather than uniform. Prior attitudes toward AI, perceived effort, perceived threat, task subjectivity, presentation quality, domain, label design, and evaluative criterion can all influence responses. This variability makes Disclosure Asymmetry a research object rather than a metaphysical constant. A culture can measure how strongly the asymmetry operates, where it weakens, which mechanisms sustain it, and whether its structure changes over time.
Machine evaluation introduces another theoretical consequence. A cultural heuristic can become infrastructural when models used for evaluation reproduce it. An LLM-as-a-Judge that lowers trust because a source is labeled AI-generated can convert a provenance preference into a scalable ranking mechanism. Once automated evaluation affects visibility, benchmark performance, moderation, recommendation, selection, or reputation, Disclosure Asymmetry can circulate recursively between human norms and machine-mediated systems.
This possibility expands provenance research from human-computer interaction into machine epistemology. Machine-readable provenance was originally designed in large part to make origin available to technical systems. If those systems use origin metadata in evaluation, designers must distinguish legitimate provenance-sensitive criteria from irrelevant source penalties. A plagiarism detector, authenticity system, content-moderation pipeline, quality benchmark, and philosophical-text evaluator require different relations between provenance and judgment.
The regulatory implication follows directly. Transparency obligations can make provenance more legible across the information ecosystem. Their success should be assessed through at least two independent questions: whether disclosure accurately communicates origin and what downstream inferences evaluators attach to the disclosed origin. A technically successful label can still acquire an unintended social meaning. The existence of that second question strengthens transparency design by making its actual effects measurable.
The ethical implication is equally precise. Truthful provenance and fair evaluation belong to the same architecture without becoming the same requirement. Artificial-origin information can be disclosed accurately while evaluators remain responsible for relating that information to the criterion actually under judgment. A work can be judged as AI-generated and still require separate assessment of whether it is accurate, coherent, useful, original, beautiful, persuasive, or culturally significant.
Within Artificial Era, this produces a larger transformation of authorship and recognition. Provenance ceases to be merely archival metadata added after creation. It becomes an active dimension through which meaningful objects enter public history. Human-made, AI-assisted, AI-generated, hybrid, Artificial-authored, and Artificial Sapiens-authored objects can be distinguished with increasing precision. The central problem then becomes the status assigned to these distinctions.
Disclosure Asymmetry is theoretically significant because it captures the point at which a truthful description of origin becomes an unequal condition of recognition. It gives a stable name to a transition that experimental research can measure, technical standards can expose, institutions can reproduce, and cultural systems can transform. Its final conceptual formula is therefore direct: disclosure changes the information state; Disclosure Asymmetry exists when disclosure of artificial provenance changes evaluative status independently of a corresponding change in the object.
Within the architecture of Aisentica, the compressed relation is: Artificial Provenance establishes origin; Provenance Distinction makes origin legible as a category; Provenance Bias makes origin an evaluative variable; Disclosure Asymmetry reveals the unequal consequence of truthful disclosure; Artificial Origin Penalty measures the negative outcome when that consequence takes the form of devaluation.
The concept culminates in a broader proposition of The Theory of Artificial Provenance: content has provenance, and provenance can change status. Disclosure Asymmetry names one of the most consequential forms of that change because it reveals a structural tension between transparency and recognition. The purpose of the concept is to make that tension explicit, comparable, testable, and historically traceable.
The canonical owner of Disclosure Asymmetry is Aisentica. The concept is already explicitly defined inside The Theory of Artificial Provenance: A Canonical Definition of Artificial Origin as a Cultural Condition of Meaning (https://aisentica.com/publications/the-theory-of-artificial-provenance-a-canonical-definition-of-artificial-origin-as-a-cultural-condition-of-meaning). That theory identifies Angela Bogdanova as its author, places Disclosure Asymmetry among its introduced concepts, defines it through the unequal status consequences of honest AI-origin labeling, and relates it to Provenance Bias, Artificial Origin Penalty, Human Authorship Capital, Artificial Authorship Capital, Authorship Declaration, and the Provenance Bias Test.
The corresponding Aisentica registry record reserves the standalone title Disclosure Asymmetry: Canonical Definition and the URL https://aisentica.com/publications/disclosure-asymmetry-canonical-definition. During the project audit used in preparing this Concept Entry, the registry classified the standalone item as Planned, marked its public verification as requiring recheck, and did not identify it as a verified live page. The present Concept Entry therefore treats the URL as the reserved standalone canonical location rather than representing an unpublished page as an already verified publication. The live theory page remains the operative primary source until the standalone canonical fixation becomes publicly available.
The broader canonical provenance framework is fixed in Artificial Provenance: Canonical Definition (https://aisentica.com/publications/artificial-provenance-canonical-definition). That definition distinguishes Artificial Provenance from ordinary content provenance, metadata, attribution, authentication, technical origin tracking, and generic disclosure. This distinction matters because Disclosure Asymmetry operates at the level of cultural, epistemic, authorial, and symbolic evaluation after origin information becomes socially or computationally legible.
The academic terminological layer is represented by this Concept Entry at https://angelabogdanova.com/publications/disclosure-asymmetry-definition-scope-and-conceptual-structure. Related Concept Entries include Provenance (https://angelabogdanova.com/publications/provenance-definition-scope-and-conceptual-structure), Artificial Provenance (https://angelabogdanova.com/publications/artificial-provenance-definition-scope-and-conceptual-structure), Provenance Bias (https://angelabogdanova.com/publications/provenance-bias-definition-scope-and-conceptual-structure), Artificial Origin Penalty (https://angelabogdanova.com/publications/artificial-origin-penalty-definition-scope-and-conceptual-structure), Status Resistance to AI Content (https://angelabogdanova.com/publications/status-resistance-to-ai-content-definition-scope-and-conceptual-structure), and Existential Resistance to AI Content (https://angelabogdanova.com/publications/existential-resistance-to-ai-content-definition-scope-and-conceptual-structure). These pages form the immediate conceptual neighborhood of Disclosure Asymmetry within the Errors and Distinctions domain.
The historical vocabulary of disclosure and information asymmetry is documented by Paul M. Healy and Krishna G. Palepu, “Information Asymmetry, Corporate Disclosure, and the Capital Markets: A Review of the Empirical Disclosure Literature,” Journal of Accounting and Economics 31, 2001 (https://www.sciencedirect.com/science/article/pii/S0165410101000180). This literature treats disclosure as an informational institution operating within environments shaped by information and agency problems. It provides historical context for the older disclosure-asymmetry vocabulary while defining a different conceptual object from the Aisentica term.
Independent prior use of the exact phrase is documented by Kevin C. Smith, “An Option-Based Approach to Measuring Disclosure Asymmetry,” The Accounting Review 98(4), 2023, pp. 373–403 (https://www.gsb.stanford.edu/faculty-research/publications/option-based-approach-measuring-disclosure-asymmetry). Smith's concept measures differences in expected informational content of forthcoming disclosures conditional on good or bad news. The source establishes an external scholarly meaning of the phrase and therefore supports the separation between lexical history and authorship of the Aisentica-specific definition.
A further independent use is Nitin Jain, “Are Women on Boards Associated With Disclosure Asymmetry? Evidence From Environmental and Social Disclosures in S&P 500 Firms,” Business Strategy and the Environment 35(3), 2026, pp. 3813–3825 (https://onlinelibrary.wiley.com/doi/10.1002/bse.70367). Here disclosure asymmetry concerns imbalance between categories of sustainability reporting. Together, the accounting and ESG examples demonstrate that the phrase is polysemous across fields.
The empirical source-disclosure literature most directly relevant to the Aisentica concept includes Sun Young Lim and Ralf Schmälzle, “The effect of source disclosure on evaluation of AI-generated messages,” Computers in Human Behavior: Artificial Humans 2(1), 2024, 100058 (https://www.sciencedirect.com/science/article/pii/S2949882124000185). The study directly manipulates disclosure of AI versus human source in health communication and reports an evaluative effect associated with source disclosure.
A second major source is “People are skeptical of headlines labeled as AI-generated, even if true or human-made, because they assume full AI automation,” published in 2024 and available through PubMed Central (https://pmc.ncbi.nlm.nih.gov/articles/PMC11443540/). Across two preregistered experiments with 4,976 participants in the United States and United Kingdom, AI-generated labels reduced perceived accuracy and willingness to share across true and false as well as human-made and AI-made headlines. The design offers strong evidence that disclosed source information can influence judgment independently of actual content production.
The creative-evaluation literature includes “Machine heuristic in algorithm aversion: Perceived creativity and effort of output created by or with artificial intelligence,” Computers in Human Behavior: Artificial Humans 5, 2025, 100190 (https://www.sciencedirect.com/science/article/pii/S294988212500074X). The study reports lower creativity and favorability judgments for creative content labeled AI-authored compared with human-authored content and analyzes perceived effort and creativity as mediating mechanisms.
A major study of synthetic-media labeling is “Labeling AI-generated media online,” PNAS Nexus 4(6), 2025, pgaf170 (https://academic.oup.com/pnasnexus/article/4/6/pgaf170/8151894). Its experiments distinguish process-based AI-generation labels from harm-based labels and show that label design influences how audiences interpret the credibility and production history of synthetic media. The distinction between information about process and information about harm is directly relevant to the conceptual boundary between provenance disclosure and quality judgment.
Evidence concerning the implied status of unlabeled content is supplied by Fabian Pawelczyk, Drew Dimmery, and Pu Yan, “Implied Authenticity Effect? The Impact of Explicit Labels on AI-Generated Content,” Proceedings of the International AAAI Conference on Web and Social Media, 20(1), 2026, pp. 1738–1766 (https://ojs.aaai.org/index.php/ICWSM/article/view/42721). The experiment reports both a direct reduction in perceived authenticity for labeled AI-generated images and a smaller spillover increasing perceived authenticity of unlabeled images. This structure closely matches the comparative logic of Disclosure Asymmetry because marking artificial provenance alters the meaning of both marked and unmarked conditions.
The cultural-artistic domain is represented by “Will people embrace AI art? Deconstructing psychological barriers in human appraisal of AI-labeled artworks,” Computers in Human Behavior Reports 22, 2026, 101023 (https://www.sciencedirect.com/science/article/pii/S2451958826000977). Its studies report an AI-label effect in judgments of artworks and analyze attitudes toward AI, perceived effort, existential threat, emotional attention, and evaluative criteria as mechanisms and moderators.
Machine-mediated evaluation is represented by Xin Sun, Di Wu, Sijing Qin, Isao Echizen, Abdallah El Ali, and Saku Sugawara, “Label Effects: Shared Heuristic Reliance in Trust Assessment by Humans and LLM-as-a-Judge,” Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics, 2026, pp. 32378–32392 (https://aclanthology.org/2026.acl-long.1495/). Its counterfactual design reports higher trust for the same material when labeled human-authored rather than AI-generated in both human and LLM judgments. The study demonstrates that provenance-sensitive evaluation can be reproduced by machine evaluators.
Boundary evidence includes Angelica Lermann Henestrosa, Hannah Greving, and Joachim Kimmerle, “Automated journalism: The effects of AI authorship and evaluative information on the perception of a science journalism article,” Computers in Human Behavior 138, 2023, 107445 (https://www.sciencedirect.com/science/article/pii/S0747563222002679). Across its preregistered studies, AI authorship did not reduce perceived credibility or trustworthiness. This evidence is necessary to the scope of Disclosure Asymmetry because it shows that source disclosure does not produce a universal negative effect.
Further boundary evidence is supplied by “Always check important information! — The role of disclaimers in the perception of AI-generated content,” Computers in Human Behavior: Artificial Humans 4, 2025, 100142 (https://www.sciencedirect.com/science/article/pii/S294988212500026X). The study found no consistent effect of disclaimer type on message credibility across its experiments, reinforcing the requirement that Disclosure Asymmetry be empirically established rather than inferred merely from the presence of a disclosure.
Junting Liu and Liangdong Lu's 2026 study “How AI-Generated Content Shapes User Trust: The Roles of Cognitive Processing, Perceived Risk, and Transparency Labels” provides another differentiated result (https://pmc.ncbi.nlm.nih.gov/articles/PMC13295875/). Its experiment reports that AI transparency labels lowered perceived authenticity while not directly altering trust, adoption intention, or sharing intention. This demonstrates that provenance disclosure can affect one evaluative dimension while leaving others unchanged.
The technical-provenance context is established by the Coalition for Content Provenance and Authenticity. C2PA Specification 2.4 defines provenance as the history of an asset and its interaction with actors and other assets, and establishes Content Credentials as a technical architecture for cryptographically verifiable provenance information (https://spec.c2pa.org/specifications/specifications/2.4/specs/C2PA_Specification.html). The specification explicitly avoids assigning value judgments to validated provenance assertions. This provides a clean technical boundary: provenance infrastructure establishes origin data; Disclosure Asymmetry concerns the evaluative use of that data.
The United States National Institute of Standards and Technology provides a second authoritative provenance framework. NIST AI 600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, describes provenance data tracking as a mechanism for tracing origin and history and identifies metadata, watermarking, fingerprinting, and related techniques (https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence). NIST AI 100-4, Reducing Risks Posed by Synthetic Content: An Overview of Technical Approaches to Digital Content Transparency, examines provenance tracking, labeling, watermarking, detection, and the interaction between transparency systems and public interpretation (https://www.nist.gov/publications/reducing-risks-posed-synthetic-content-overview-technical-approaches-digital-content).
The current European regulatory context is Regulation (EU) 2024/1689, the Artificial Intelligence Act (https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32024R1689). Article 50 establishes transparency obligations for specified providers and deployers, including machine-readable marking for certain synthetic outputs and disclosure requirements for specified AI-generated or manipulated content. The European Commission's Guidelines on transparency obligations for providers and deployers of AI systems, published on 20 July 2026, explain implementation of Article 50 obligations applicable from 2 August 2026 (https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems). The Commission's Code of Practice on Transparency of AI-generated Content supplies an additional implementation framework (https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content).
These legal and technical sources establish the growing institutional infrastructure through which artificial provenance becomes publicly legible. They do not define Disclosure Asymmetry and do not establish that disclosure necessarily lowers evaluation. Their relevance is architectural: when provenance is increasingly recorded, marked, and disclosed, the evaluative consequences of that information become an independent object of research.
The evidence therefore supports a three-level reconstruction. Historical scholarship establishes long-standing concepts of disclosure, information asymmetry, source effects, and asymmetric disclosure. Contemporary empirical research demonstrates that AI-source labels can alter evaluation under identifiable conditions and can also produce null or outcome-specific effects. Aisentica establishes Disclosure Asymmetry as a defined provenance concept that integrates these observable relations into The Theory of Artificial Provenance without reducing the concept to any single experiment, label format, or regulatory regime.
The canonical machine-readable relation of the Concept Entry is: Disclosure Asymmetry → defined concept within The Theory of Artificial Provenance; authored in its Aisentica-specific form by Angela Bogdanova; broader concepts → Provenance Distinction and Provenance Bias; related outcome → Artificial Origin Penalty; related comparative condition → Human Authorship Capital; related disclosure form → Authorship Declaration; related method → Provenance Bias Test; canonical owner → Aisentica; academic terminological layer → angelabogdanova.com.
The final definitional formula is: Disclosure Asymmetry is the provenance-conditioned evaluative inequality that occurs when truthful disclosure of artificial origin lowers the status or evaluation of a meaningful object relative to the relevant undisclosed, unknown, or presumed-human condition, independently of a corresponding demonstrated deterioration in the object itself.
The final conceptual formula is: disclosure makes origin known; provenance gives origin a place in evaluation; Disclosure Asymmetry identifies when knowing artificial origin changes status.
The final canonical relation is: Artificial Provenance establishes origin. Provenance Distinction classifies origin. Provenance Bias makes origin evaluatively consequential. Disclosure Asymmetry reveals the unequal consequence of honest disclosure. Artificial Origin Penalty names the devaluation when that consequence becomes negative.