*
Correspondence. Ricardo F. Morín. CGI, 2026.
Ricardo F. Morín
November 2, 2026
Oakland Park, Fl
A viewer who treats a production label as sufficient evidence of a creator’s competence substitutes information about a tool for an examination of the creator’s decisions. The platform describes a process; the viewer infers how much the creator delegated, assigns a value to that delegation, and credits or discounts the creator accordingly. The viewer’s judgment therefore includes assumptions about human conduct that the platform’s description of the process has not supplied.
A designer reports creating a graphic in Canva for her employer’s social media account, using neither a template nor artificial intelligence. After publication, she encounters an AI label that she interprets as an assertion that a machine produced the entire work. She protests that she designed the composition and describes her predicament as having to convince a robot that she is not a robot. She then attributes laziness to the convenience of AI, describes business use of AI as “cheap and lazy,” and claims that graphic designers possess an authenticity that AI cannot supply. She asks readers to support graphic designers and recommends Canva Pro. She also proposes taking a screenshot before publication to prevent the platform from reading identifying metadata.1
The designer disputes the platform’s account of her process and interprets the notice as a denial of her contribution. Verifying her objection to the process description would require a record of the operations she used and the wording of the notice. Assessing her contribution would require attention to the decisions she made during production. A platform can correctly identify an AI operation while omitting the purposes and choices through which a person directed that operation. A reader may take that limited description as a complete account of the composition and credit the system with decisions that the notice has not documented.
A designer exercises agency by directing an undertaking and deciding how to respond to the results. The designer claims authorship for the contribution those decisions make to the work. Software providers determine which operations an interface permits, including operations performed through AI. Within those limits, the designer can assess a proposed composition, refuse it, or direct a revision. The designer exercises autonomy through those choices, even while relying on the instrument to execute them. An observer assessing authorship needs to distinguish the choices the designer exercised from the operations delegated to the instrument.
A user working through familiar software may describe the process through the actions the user remembers performing: selecting an element, changing its position, or rejecting an arrangement. The account gives prominence to the user’s interventions and less attention to the operations the software executed. A viewer reading an AI label may reverse that emphasis, crediting the system with the result while overlooking the interventions of its user. The two accounts direct attention toward different participants in the same activity. Familiarity with an interface helps explain the user’s description, but familiarity cannot measure the contribution of another person working through an unfamiliar interface.
The designer identifies Canva as her working platform and reports that she used no AI. Canva describes its Background Remover as using AI while classifying the feature as an editing tool that does not add AI-generated-content metadata to a design.2 The provider thus distinguishes the technology used by a feature from the category recorded in the resulting file. A user can encounter an operation as ordinary editing while the provider identifies AI within its execution. The designer’s identification of Canva names the platform; her claim to authorship concerns what she decided while using it.
The designer asks readers to recognize her own decisions, yet her description of AI use as “cheap and lazy” discounts other creators without examining their decisions. She invokes attention to a particular contribution when defending her work and substitutes a general judgment about a tool when assessing work by others. The platform’s possible factual error and her unfavorable judgment still require different corrections: the platform would have to correct an inaccurate process description, while the designer would have to supply grounds for her assessment of the people using the process.
A business could commission a notice explaining that its entrance has moved to a side street during building repairs. A designer could obtain an attractive image through an automated process, recognize that the image directs customers toward the closed entrance, and require a correction. Another designer could construct every visible element through separately controlled operations and nevertheless retain the misleading entrance. The first designer has acted on information essential to the commission. The second designer has overlooked that information despite controlling the individual operations.
The client can examine how each designer used the information about the entrance and whether the completed notice directs customers to the correct door. The client may also judge the arrangement of the image and text, which presents a different question from the accuracy of the directions. One designer could supply correct information in an awkward composition; another could organize the composition with invention and omit an essential instruction. The two assessments address different decisions by the designer: which directions to give and how to arrange the visible elements.
Drafts and correspondence can record decisions that the finished work conceals. A rejected image may document a mistake the designer caught before delivery. An explanation accompanying a revision may connect the change to a concern the client expressed. Such records allow the client to compare the designer’s account with the development of the work. A persuasive account may exaggerate the contribution, while a capable practitioner may struggle to explain judgments exercised through habit or experience. Neither verbal fluency nor difficulty in explanation provides a sufficient account of the work.
An observer unfamiliar with the production process may overlook decisions that another practitioner would recognize. A report of finding no evidence of judgment does not, by itself, determine whether the observer missed relevant decisions or whether the creator exercised no judgment over the aspect being assessed. Professional standing also gives no account of the particular decisions under review; a favorable reputation can encourage confidence without resolving a question about the work.
The designer in the complaint reports publishing through her employer’s account. Her employer relies on the work she supplies, while the platform presents its own classification to the audience. She may control the composition without controlling the notice attached to its publication. Viewers can then use the platform’s description to assess a contribution whose development they never witnessed. Her authority over the work and the platform’s authority over its presentation therefore concern different parts of the same professional relationship.
In commissioned work, the client requests an image, the designer develops its composition, and the publisher decides whether and how to display it. Each participant determines a different part of what reaches the audience. Approval of a finished image documents the client’s acceptance; approval alone does not document authorship of the arrangement. A viewer may assign every consequential choice to a single author, including choices the client or publisher made.
A client might read an AI label and infer that software determined the entire composition. If the client then equates that assumed delegation with negligible effort, the client may question the fee or discount the designer’s competence. The label names a production process. The client adds conclusions about the extent of the delegation, the effort involved, and the value of the service. Unless the client examines the designer’s actual decisions, the client’s assessment rests on those added assumptions.
In July 2024, Meta acknowledged that its systems had attached a “Made with AI” label to some images whose creators had made only minor AI modifications, including retouching. Meta attributed the labels to indicators included in those files, acknowledged that its wording had not always provided enough context, and announced a change to “AI info.”3 The company had identified the participation of a technology, but its notice had not adequately conveyed the extent of that participation.
The Coalition for Content Provenance and Authenticity describes Content Credentials as records containing claims about a digital asset’s origin and subsequent handling. The coalition’s specifications distinguish the verification of those records from an assessment of whether the depicted content is true, accurate, or factual. The coalition also cautions readers against judging trustworthiness solely from the presence or absence of credentials.4 A viewer may use verification as grounds for trusting the creator. The verification procedures check the record’s integrity; the viewer supplies the judgment about the creator’s sincerity or creative worth.
An employer may consult a production record to check compliance with a stipulated method. A critic may examine the composition to assess how the arrangement presents its subject. The employer and the critic maintain relational coherence when each supports the judgment with evidence appropriate to the responsibility undertaken: enforcing a stated condition of the work or interpreting the composition.
A platform that places a notice beside an image presents the classification at the moment the viewer encounters the work. The creator’s later explanation may reach the viewer only if the viewer seeks it out or returns to the discussion. By controlling the placement of the notice, the platform can give its account priority over the creator’s account before anyone has compared the two. The creator then has to secure attention for a correction and provide reasons for reconsidering the first impression.
Platform operators choose the categories their systems apply and the procedures through which creators can contest a classification. Developers implement those choices. An automated notice can therefore apply an institutional decision to an image that no operator has individually examined. Viewers decide how much authority to grant the resulting notice when assessing the creator’s contribution.
The designer’s screenshot proposal addresses the notice by attempting to withhold the information that might trigger it. If her proposal succeeded, the platform would omit the label while presenting substantially the same composition. She would have changed the conditions under which viewers encountered the work without revising the decisions that produced its arrangement. Viewers could credit her with more judgment after the label disappeared, even though she had made no additional decisions about the original composition.
In a different case, a creator might offer an image as a record of an event. Applying the designer’s proposed workaround to that image could remove information relevant to the documentary claim. Viewers would then have less information with which to check whether the creator generated the scene they were being asked to accept as a record of something that happened.
The designer also asks readers to pay graphic designers to produce flyers, logos, and social media posts. She invokes authenticity as a reason to purchase those design services. A prospective client may understand that claim to promise knowledge of the business, a distinctive composition, a particular method, or access to someone who will correct mistakes. The word authentic does not specify which of these contributions the designer promises. A client could therefore purchase the service on an understanding the designer has not expressly confirmed.
If a client commissions a drawing executed in pencil on paper, a designer who substitutes a generated image has changed an agreed condition. The client may admire the delivered image and still object to the substitution. Through that objection, the client invokes the promised medium; the objection gives no account of the artistic value of images produced under other agreements.
When a client leaves the choice of tools open, the designer still undertakes to fulfill the agreed purpose and constraints. If the chosen process prevents a revision the designer has promised, the choice of method has limited the service despite the absence of a stipulation about tools. By describing the process, the designer gives the client an opportunity to compare its limits with the expected service. The participants preserve transactional coherence when the designer’s representations, the client’s understanding, and the supplied service correspond. A later objection to a tool may identify a departure from the agreement or introduce a condition the designer never accepted.
A designer familiar with a business may recognize a reference that would confuse its customers or understand why the business retains a familiar visual element. The designer may also respond to a request for correction before a public event. A client can value the designer’s knowledge of the business and responsiveness while accepting the designer’s use of AI or other software.
A client might recognize the care a designer supplies and still choose a cheaper service that the client considers adequate. The designer directs the execution of the work, while the client controls whether to purchase that service again. A client can also accept a correction to a production label and acknowledge the designer’s contribution while declining to offer another commission.
The designer’s concern about displacement therefore reaches beyond attribution. If she makes her appeal for support depend on the universal inferiority of AI-assisted work, a client who encounters competent work can reject that ground for support without addressing the loss of income. In that situation, the client has answered her claim about a competing method; she still has to sustain a livelihood under the changed demand.
A platform can automate the classification of a file. A reader may then accept the classification as a complete account of the creator’s contribution and judge the creator’s competence or professional value on that basis. The reader has supplied the connection between the described process and the verdict on the person. The platform answers for its description; the reader answers for the conclusions drawn from it.
Notes
2. Canva, “Remove photo and video backgrounds,” Help Center. The page identifies Background Remover as an AI editing feature and states that the feature does not add Canva’s AI-generated-content metadata. That description concerns the feature; the designer’s account does not identify every operation she used. Consulted November 2, 2026.
https://www.canva.com/help/background-remover/
3. Meta, “Our Approach to Labeling AI-Generated Content and Manipulated Media,” originally published April 5, 2024, with the July 1, 2024 update explaining the change from “Made with AI” to “AI info.” Meta’s historical acknowledgment documents a limitation of its labeling practice; it does not determine what triggered the particular notice disputed by the designer. Consulted November 2, 2026.
https://about.fb.com/news/2024/04/metas-approach-to-labeling-ai-generated-content-and-manipulated-media/
4. Coalition for Content Provenance and Authenticity, “C2PA and Content Credentials Explainer,” version 2.2. Section 2 describes the scope of provenance verification; section 7.1.2 addresses trust and missing credentials; section 7.2.2 distinguishes provenance from factual truth. Consulted November 2, 2026.
https://spec.c2pa.org/specifications/specifications/2.2/explainer/Explainer.html
