Understand the technology
From clear disclosure to informed decisions.
For May and November 2027. This lesson is for SL and HL students. Learn through studio examples, disclosure comparisons and a transparent delivery workflow.
01 | A BEAUTIFUL IMAGE CAN STILL CREATE A TRUST PROBLEM
Imagine Visionary Studios sends a client a polished campaign image. The client approves it believing the scene was photographed, then discovers that the people and location were generated. The image may meet the visual brief, but the client made a decision without information they considered important.
Transparency means making relevant information understandable and available to the people who need it. In this lesson, that includes the role of AI, important limitations and the checks performed.
This scenario is a classroom illustration, not an event reported in the official case study.
02 | DISCLOSURE IS A COMMUNICATION DECISION
AI disclosure is communicating that AI contributed to creating or modifying material. Useful disclosure explains the contribution rather than relying on an unexplained label.
Compare “AI involved” with “The background was generated; the product photograph was supplied by the client.” The second tells the client what changed and helps them ask appropriate questions.
Decide what information matters to the intended decision. A client choosing whether to publish an asset may need different detail from a viewer scrolling past the final advertisement.
03 | WHO NEEDS TO KNOW WHAT?
| Audience | Useful information | Decision supported |
|---|---|---|
| Client | AI contribution, limitations, review status and unresolved concerns | Whether to approve and how to use an asset |
| Internal reviewer | Source records, tool version, edits and test evidence | Whether requirements and checks are met |
| Public audience | Clear relevant information about generated or altered content | How to interpret what is shown |
Transparency does not mean sending every internal record to everyone. It means communicating enough appropriate information without exposing confidential material.
04 | AGREE THE WORKFLOW BEFORE PRODUCTION
Discuss proposed AI use when agreeing the creative brief. Explain which stages may use generation, what human work will follow and how approval will operate.
This gives the client a chance to raise concerns before the studio invests in a workflow. A late surprise can cause rework even when the studio believed the tool was suitable.
Record the agreed approach and revisit it if the contribution changes. Approval for a generated background should not silently become approval for generated product features.
05 | STATE WHAT AI DID
AI can support brainstorming, generate a background, alter an existing image or produce most of the final scene. These contributions are not equivalent.
A useful description distinguishes the generated parts from client-supplied material and human editing. It also avoids suggesting that a real event was photographed when it was not.
For Visionary Studios, an asset record might say: “Background generated from a text brief; product photograph composited by a designer; colour and layout edited manually.”
06 | STATE WHAT IS KNOWN AND UNKNOWN
A studio may know its prompt and edits but lack full visibility into a provider’s training data. Transparency includes acknowledging that boundary.
Do not turn a provider’s general assurance into a stronger studio claim. “We reviewed the provider’s published information” differs from “We independently verified every training image.”
Clear uncertainty helps clients make informed choices. It is more useful than either hiding a gap or making an unsupported guarantee.
07 | GENERATED SCENES ARE NOT EVIDENCE OF EVENTS
A realistic image can look like a record of something that happened. A generated crowd, building or product demonstration does not establish that the event or performance occurred.
Consider a fictional advertisement showing a product in an impossible environment. The intended interpretation may be obvious. A realistic image presented as documentary evidence creates a different risk of misleading the viewer.
Evaluate the context, wording and placement together. A technically accurate disclosure hidden far from the image may not correct the main impression.
08 | MAKE DISCLOSURE CLEAR AND TIMELY
Useful wording is specific, readable and placed where the intended audience is likely to encounter it before making the relevant decision.
“AI-generated concept image; not a photograph of a completed building” tells a prospective client more than a tiny unexplained symbol. The wording should match the actual contribution.
Do not assume that adding any label anywhere is sufficient communication. Ask a reviewer unfamiliar with the project what they understand from the image and accompanying text.
09 | THREE DISCLOSURES TO COMPARE
| Statement | Assessment |
|---|---|
| “100% verified and risk-free.” | An absolute claim requiring evidence the workflow is unlikely to establish. |
| “Created using technology.” | Too vague to explain the relevant AI contribution. |
| “The background is AI-generated. The product image is a client-supplied photograph, composited by our design team.” | Specific about contribution, provided the statement is accurate. |
The strongest wording still does not prove rights clearance, fairness or factual accuracy. Disclosure is one part of responsible production.
10 | PROVENANCE SUPPORTS AN EXPLANATION
Provenance is information about an asset’s origin and processing history. Record the source materials, tool and version where available, relevant settings, major edits and review decisions.
These records help the studio answer questions later. If a client asks which parts were generated, the team should not have to rely on someone’s memory.
Records support accountability, but they can be incomplete or inaccurate. Having a log is not proof that every statement in it is correct.
11 | LABELS, METADATA AND WATERMARKS HAVE LIMITS
A visible label can inform viewers directly. Metadata can carry information within a file, while a watermark can signal origin or processing. Their usefulness depends on the implementation and the audience.
Metadata can be lost when files are exported or shared. A visible mark can be cropped or overlooked. A provenance mechanism may support a history claim without proving that the depicted scene is true.
Use appropriate complementary methods rather than treating one technical signal as a guarantee of authenticity.
12 | HUMAN REVIEW MUST MEAN SOMETHING
“Human reviewed” is useful only if the studio can explain what was checked. A designer approving colour balance has not necessarily assessed cultural accuracy, rights or product claims.
Define review responsibilities and record the checks completed. Separate an approval for visual quality from a decision about unresolved concerns.
This prevents a general phrase from creating an impression of more extensive assurance than the process provides.
13 | TRANSPARENCY AND CONFIDENTIALITY CAN COEXIST
The studio may hold unreleased product images, client plans or sensitive prompts. Being transparent does not require publishing those materials.
Provide appropriate summaries and restrict detailed records to people who need access. For example, describe the generated contribution publicly while keeping an unreleased client brief within the project team.
The practical question is how to explain the workflow honestly while protecting information entrusted to the studio.
14 | TRUST IS BUILT BY CONSISTENT BEHAVIOUR
Trust is confidence that another party will act competently and honestly. Clear disclosure can support it, but a label alone does not create it.
Clients also notice whether the studio meets agreed requirements, answers questions accurately and responds well when something goes wrong. Repeatedly overstating certainty can undermine confidence even if an individual asset looks excellent.
Transparency should therefore be a working practice across briefing, review and delivery, not a sentence added only at the end.
15 | RESPOND WHEN A PROBLEM IS DISCOVERED
Suppose the team discovers that a delivery note incorrectly described a generated scene as a photograph. Preserve the relevant records, investigate the error and inform the appropriate decision-makers promptly.
Explain what is known, which assets are affected and what correction is proposed. Update inaccurate statements and review how the mistake entered the workflow.
Trying to conceal the issue can create further harm. A clear response supports accountability, although it cannot guarantee that trust will immediately recover.
16 | TEST THE COMMUNICATION
Show a draft image and disclosure to people unfamiliar with production. Ask what they believe is real, generated or edited, and what they think “reviewed” means.
In a fictional test, 9 of 20 readers correctly identified the generated background with vague wording: 45%. With specific wording, 17 of 20 did: 85%. The difference is 40 percentage points .
This suggests clearer understanding in that small classroom test. It does not establish universal comprehension or prove that the image is accurate. Review remaining misunderstandings and test with relevant audiences.
17 | EVALUATE THE COSTS AND BENEFITS
Maintaining records, explaining limitations and reviewing wording take time. Clients may ask additional questions, and disclosure can affect how an audience interprets an image.
However, clear communication can reduce surprises, improve approval decisions and help the studio address concerns. Hiding material information may produce short-term convenience at the expense of later confidence.
Evaluate the actual context and evidence. Do not claim that disclosure always increases trust, or that the fear of questions justifies misleading communication.
18 | BUILD A TRANSPARENT DELIVERY WORKFLOW
Discuss AI use and approval expectations.
Document contributions, checks and uncertainties.
Deliver clear information and respond to concerns.
In your book, propose a delivery note for a campaign containing a generated background and a real product photograph. Include the contribution, scope of review and any unresolved issue that affects approval.
Reveal a possible response
“The background is AI-generated; the product photograph was supplied by the client and composited by our design team. We checked visual consistency and the product’s appearance against the supplied reference. The generated location is fictional. The identified outstanding concern is recorded separately for client decision before publication.”
The wording must reflect checks actually completed. Use these independent teaching notes alongside the official 2027 booklet; examples on this page are fictional. This lesson discusses communication and ethics rather than setting out jurisdiction-specific disclosure rules.
CHECK YOUR UNDERSTANDING | CLASSIFY THE TRANSPARENCY ACTION