Every AI agency will face this moment: a prospect or client asks, directly or indirectly, whether AI was involved in producing the work. How you answer — and whether you were proactive about it — shapes how they perceive your integrity, your quality, and their ability to trust what you deliver.
The instinct to hide AI involvement is understandable. Clients have uneven relationships with AI-generated work. Some are enthusiastic. Some are suspicious. Some have explicit prohibitions in their brand guidelines. Proactive disclosure risks losing a deal to a client who would have been fine with AI if they'd never been told about it. Reactive disclosure — waiting until asked — feels safer.
But proactive disclosure has a different logic. Clients who discover AI use after the fact — especially if they paid for what they thought was human work — are far more damaged relationships than clients who were told upfront. And under the EU AI Act, certain AI-generated content carries mandatory transparency obligations. The question is not whether to disclose. It's when, how, and to whom.
An AI content agency has three active clients in the same month. Each one surfaces the disclosure question differently.
Client A is a B2B software company. They hired the agency to produce ten blog posts per month. The brief says "expert-quality content." Nothing in the contract mentions AI. The client's marketing director has never asked about the production process. The posts are going live with the marketing director's name as the listed author. The posts are well-written — the agency uses AI drafting with human editing and fact-checking. No errors have been found. The client is happy. No one has asked.
Client B is a regulated financial services firm. They hired the agency for compliance communications — internal policy summaries for employees. The contract is silent on AI. But the firm is subject to EU financial services regulations that require disclosed, auditable production processes for compliance materials. The client's legal team has never reviewed the contract. The agency's AI drafts are reviewed by a human before delivery. The review is thorough. But the client doesn't know an AI was involved at any stage.
Client C is a personal brand account — a consultant who sells their expertise as their product. They hired the agency to ghostwrite their LinkedIn posts and newsletter. The client explicitly said they want their "authentic voice." The agency uses AI to produce drafts that the client edits and approves before publishing. The client knows AI is in the workflow — they specifically requested it. But they've asked the agency not to tell anyone, including their audience.
Three clients. Three transparency situations. Different stakes, different legal exposure, different reputational risks.
This module asks: what's your disclosure policy, and how does it apply to each of these clients?
Disclosure is not purely a strategic choice. Part of it is a legal requirement — and the requirements are growing. Understanding what the law mandates helps you distinguish between the disclosure decisions that are required and the ones that are yours to make.
Article 13 requires providers of high-risk AI systems to ensure their systems are transparent — meaning the deployer (the business using the AI) can understand the system's capabilities, limitations, and conditions of use. For AI agencies, this means the client must be able to understand what the AI can and cannot do in the context of the work they're receiving. This is a deployer-level obligation, not necessarily an end-user disclosure. It applies whether or not the final audience knows AI was involved.
Separately, the EU AI Act requires that AI-generated images, audio, and video that could deceive people as to their authenticity must be labeled. This applies to AI-generated text in some specific contexts — political advertising, for instance, has its own disclosure requirements under EU regulation. For most B2B agency work, this specific clause has limited direct application. But the underlying principle is clear: when AI-generated content could be mistaken for human-produced content in a context where that distinction matters, disclosure is required.
Article 13 requires that high-risk AI systems be designed and developed so that their operation is sufficiently transparent to enable deployers to understand and use the system's outputs appropriately, and to implement human oversight. For AI agencies delivering work to business clients, this means the client must understand what AI was involved, what it produced, and what human review was applied. An agency that withholds this information from a client who would use it to govern their own processes may be in violation of Art.13 obligations.
UNESCO's 2021 AI Ethics Recommendation emphasizes that AI systems should be transparent — meaning stakeholders should have access to information about the AI systems that affect them, and there should be clear accountability for AI-generated outputs. For agency work in sensitive domains (financial services, healthcare, legal, or regulated industries), the UNESCO standard supports proactive disclosure as both an ethical practice and a risk management measure.
The law draws a floor. Your agency's policy sits somewhere above it. Where exactly is the debate.
When you face a disclosure decision — whether to disclose, what to disclose, and to whom — three tests cut through the ambiguity. Apply them in order.
Would a reasonable person in the client's position, knowing the type of work they hired for, expect that AI was or was not involved? A client who hired an agency for "expert content" may have a reasonable expectation of human expertise. A client who hired an agency for "AI-powered workflows" clearly does not expect human-only production. If there's a plausible reasonable expectation of no AI involvement, proactive disclosure is required — not because you're obligated to confess, but because allowing a false impression to form is a deception.
Would knowing about AI use change the client's decision to buy, their use of the output, or their legal obligations? If the answer is yes to any of these, disclosure is required regardless of what the contract says. Client B in the scenario — the regulated financial firm — has legal obligations that are affected by AI involvement in compliance materials. Their not knowing doesn't protect the agency; it exposes both parties.
If something goes wrong with the AI-generated output, and the client later discovers AI was involved without their knowledge, what is their recourse? And what is yours? Clients who were not disclosed to have a much stronger case for disputing payment, claiming breach of contract, or escalating to regulators. Disclosure upfront removes that exposure — and creates a documented record that the client understood and accepted the AI-assisted delivery model.
You'll apply all three tests in the lab — debating your disclosure position across the three client scenarios from the Scenario page.