Specifying which parts of online content were generated by AI can blunt the trust-damaging effects of blanket AI disclosures; scope-based labels make AI involvement more informative and convey seller effort, improving consumer responses.

From Compliance to Communication: How Explainable AIGC Disclosure Shapes Consumer Responses in E-Commerce

Kaixin Sha, Qi Wang, D Xu · Fetched June 16, 2026 · Journal of the Association for Information Systems

openalex rct medium evidence 7/10 relevance Source PDF

The study tests whether scope-based AIGC disclosures—that specify which part of content was AI-generated—reduce the trust- and purchase-intention harms of generic AI disclosures by increasing perceived diagnosticity and signaling seller effort.

Given the increasing prevalence of Artificial Intelligence-Generated Content (AIGC), regulatory authorities have begun requiring firms to disclose their involvement with AI to enhance transparency and prevent deception. However, emerging evidence suggests that general disclosures of AI involvement may unintentionally undermine consumer trust and reduce purchase intentions. This study examines whether and how scope-based AIGC disclosure can mitigate these adverse effects in e-commerce contexts. Drawing on cue utilization theory, we conceptualize the scope-based AIGC disclosure as an information cue that clarifies which part of the content was created by AI. We propose that scope-based AIGC disclosure shape consumer responses through two parallel mechanisms: perceived diagnosticity and perceived seller effort. We plan to conduct a lab experiment to test these arguments. This study contributes to AI transparency and explainability research by shifting attention from whether firms disclose AI use to how disclosure is designed in digital commerce.

Summary

Main Finding

This study hypothesizes that scope-based AIGC disclosure — a label that specifies which part(s) of digital content were produced by AI — can mitigate the negative effects of general AI-use disclosures on consumer trust and purchase intentions in e-commerce. Specifically, scope-based disclosure is expected to improve outcomes relative to a generic “AI was used” label by increasing (1) perceived diagnosticity of the information and (2) perceived seller effort, which jointly restore trust and willingness to buy.

Key Points

Regulatory context: Authorities increasingly require firms to disclose use of AI in consumer-facing content to increase transparency and prevent deception.
Problem: Emerging evidence suggests that generic/unspecified AI disclosures can backfire, reducing consumer trust and purchase intentions.
Conceptual innovation: Introduces scope-based AIGC disclosure as an information cue that clarifies which portion of content (e.g., product description, photos, reviews) was AI-generated rather than merely indicating AI involvement.
Theoretical framing: Uses cue-utilization theory to argue that consumers use disclosure format as a cue; clearer/specific cues (scope-based) should be more diagnostic and convey greater seller effort.
Proposed parallel mechanisms:
- Perceived diagnosticity: Scope-based labels give consumers clearer, actionable information about content provenance, improving perceived informativeness and enabling better judgments.
- Perceived seller effort: More specific disclosure signals seller care and accountability (effort to be transparent), which increases trust.
Expected net effect: Scope-based disclosure → higher perceived diagnosticity & seller effort → higher trust and purchase intentions versus generic disclosure (and possibly versus no disclosure, depending on baseline effects).

Data & Methods

Research design: Controlled lab experiment (e-commerce context) to manipulate disclosure type.
Suggested experimental conditions (implied by theory): at minimum — (a) no disclosure, (b) generic AI-use disclosure, (c) scope-based disclosure specifying which parts are AI-generated. Additional variants could manipulate level of scope granularity (e.g., single element vs multiple elements).
Sample: Human subjects (online panel or lab participants) recruited to evaluate product listings or other commerce content.
Stimuli: Simulated product pages or content examples with identical informational content but differing in disclosure text/labels.
Measures:
- Dependent variables: consumer trust, purchase intention/willingness to buy, perceived deception.
- Mediators: perceived diagnosticity of the disclosure, perceived seller effort/intentionality.
- Controls: product type, prior AI attitudes, familiarity with AI, demographic covariates.
Analyses:
- Between-subjects comparisons (ANOVA/OLS) of disclosure conditions on outcomes.
- Mediation analysis (e.g., structural equation modeling or causal mediation) to test whether perceived diagnosticity and seller effort explain treatment effects.
- Robustness checks (e.g., moderation by prior AI attitudes or product risk).
Timeline/status: Planned lab experiment (study not yet executed in the provided description).

Implications for AI Economics

Policy design: Findings would inform regulators on disclosure standards. Mandates requiring mere AI-use notification may be counterproductive; specifying scope could achieve transparency goals while avoiding demand damage.
Market outcomes: Disclosure design affects consumer trust and purchase decisions, altering demand for AIGC-enabled offerings. Firms may strategically choose disclosure formats that minimize negative demand effects while complying with regulation.
Signaling and competition: Scope-based disclosure can act as a signal of seller quality/effort; heterogeneous adoption could create competitive differentiation and influence pricing, entry, and investment in human vs AI content production.
Welfare and adoption: Better-designed disclosures could preserve consumer welfare (avoid unnecessary pessimism about AI) while maintaining protection against deception. This influences overall adoption rates of AIGC and the diffusion of AI in commerce.
Future empirical work: Results would motivate field experiments and marketplace data analysis to quantify real-world demand shifts, second-order effects (seller reputation dynamics), and optimal regulatory cost–benefit tradeoffs.

Assessment

Paper Typerct Evidence Strengthmedium — A randomized lab experiment provides strong internal validity and can credibly identify the causal effect of disclosure format on measured attitudes; however, external validity is limited by the artificial lab setting, likely convenience sampling, hypothetical purchase measures, single product/category stimuli, and unknown sample size and implementation details, which reduce confidence in generalizing to real-world e-commerce outcomes. Methods Rigormedium — The design (randomized manipulation, theory-driven mediators) is appropriate and can yield clear causal estimates if properly implemented; rating is tempered by missing methodological details in the proposal (pre-registration, power analysis, manipulation checks, stimulus realism, incentive-compatible purchase behavior, and robustness checks), and by potential measurement and mediation identification assumptions that should be addressed. SamplePlanned laboratory experiment with recruited consumers (likely a convenience or online panel sample) exposed to e-commerce product/content stimuli where disclosure type is manipulated; outcomes include trust, purchase intention, perceived diagnosticity, and perceived seller effort; exact sample size, recruitment source, demographics, and whether real incentives/purchases will be used are not specified. Themesgovernance adoption IdentificationRandom assignment of participants in a laboratory experiment to different disclosure conditions (e.g., generic AI disclosure vs. scope-based AIGC disclosure vs. control), with between-subjects manipulation of the disclosure cue; mediation analysis used to test perceived diagnosticity and perceived seller effort as causal pathways (with the usual assumptions for causal mediation). GeneralizabilityLab/experimental setting may not reflect real-world online shopping behavior, Likely convenience or student/online panel sample limits population representativeness, Single product/category and specific stimuli may not generalize across product types or industries, Short-term measured attitudes (intentions) may not predict actual long-term purchase behavior, Cultural or regulatory context (country-specific disclosure norms) may limit cross-country applicability

Claims (6)

Claim	Direction	Outcome	Confidence & Evidence	Details
Regulatory authorities have begun requiring firms to disclose their involvement with AI to enhance transparency and prevent deception. Governance And Regulation	positive	requirement for firms to disclose AI involvement (regulatory disclosure)	Reading fidelity high Study strength medium	0.6
Emerging evidence suggests that general disclosures of AI involvement may unintentionally undermine consumer trust and reduce purchase intentions. Consumer Welfare	negative	consumer trust and purchase intentions	Reading fidelity high Study strength medium	0.6
Scope-based AIGC disclosure can mitigate the adverse effects (reduced trust and purchase intentions) of general AI-use disclosures in e-commerce contexts. Consumer Welfare	positive	consumer trust and purchase intentions (mitigation of adverse effects)	Reading fidelity high Study strength speculative	0.1
Scope-based AIGC disclosure can be conceptualized, drawing on cue utilization theory, as an information cue that clarifies which part of content was created by AI. Decision Quality	positive	clarity/diagnosticity of information provided to consumers	Reading fidelity high Study strength low	0.3
Scope-based AIGC disclosure shapes consumer responses through two parallel mechanisms: perceived diagnosticity and perceived seller effort. Decision Quality	positive	perceived diagnosticity and perceived seller effort (mediators influencing consumer responses)	Reading fidelity high Study strength speculative	0.1
The authors plan to conduct a lab experiment to test the proposed effects of scope-based AIGC disclosure on consumer responses. Adoption Rate	null_result	planned experimental test of consumer responses to disclosure designs	Reading fidelity high Study strength speculative	0.1