Samuel Idris runs a finance operations team of nine at a regional logistics company. He had quietly started using an AI assistant to speed up the parts of his week he hated: drafting the monthly variance commentary, summarizing a wall of vendor emails, sketching the first version of his team update. It worked. He was getting an hour back most days. Then, in a one-on-one, his most senior analyst, Priya, asked a flat question: "Did a bot write the feedback you gave me last month?" Samuel froze. He had used AI to help organize his notes before that review. He had not lied, but he had not said anything either. He mumbled something vague, and he watched her trust in him drop a notch in real time. That conversation is why Samuel rebuilt how he talks about AI. This lesson is what he learned.

Why secrecy quietly costs you

Samuel realized the problem was not that he used AI. Using AI to draft and organize is legitimate, even smart. The problem was the silence around it. In finance, where his whole job rests on his team and his auditors believing the numbers and the person behind them, an undisclosed shortcut is not a small thing. Secrecy about AI use erodes trust; transparency builds it. When people discover later that AI was involved in something that touched them, it feels like deception even when none was intended, and it triggers exactly the fears Samuel did not want: that he was automating judgment, that his feedback was generic, that he was hiding something.

The flip side is the opportunity. When Samuel is open about how he uses AI, his team understands he is using tools responsibly, speculation dies down, expectations get clear, and he models the same honest behavior he wants from them. In a finance team, where he is also going to ask his analysts to use AI for reconciliations and reporting, that modeling matters: they will copy how their manager handles it.

The disclosure is rarely the thing that damages trust. The discovery is. If finding out later would feel like a betrayal, that is your signal to say it now.

When to disclose, and when you do not need to

Samuel built himself a simple rule of thumb so he was not deciding from scratch each time. He discloses in three situations. First, when the AI use directly affects a person: he used AI to help organize the feedback for Priya, or to draft a message to the team, or to analyze team metrics, or to prep for a meeting with someone. They should know, because if they find out later it reads as deception. Second, when he is asking for someone's involvement: teaching an analyst to use AI, asking them to check an AI output, or explaining his own workflow. People cannot consent to something they do not know about. Third, when the output goes outside the team and becomes customer-facing or public: a note to a client, a figure that lands in a board pack, anything that might later be quoted or audited. Stakeholders may care that AI was involved, and being caught hiding it is far worse than saying so up front.

He also gave himself permission not to narrate every keystroke. Private, low-stakes use that affects no one else does not need a broadcast: using AI to organize his own notes, to brainstorm for himself, or to help him understand a dataset he then verifies. The principle is clean: if it only affects you and you reviewed it, you do not need to announce it, but if someone asks, you answer honestly. Over-disclosing trivial use ("I used AI to help structure my meeting agenda") just adds noise and makes you look anxious about a legitimate tool.

How to disclose well

Samuel learned that how he says it matters as much as whether he says it. Vague disclosure is almost as bad as silence. "I used AI for this" tells people nothing and can actually increase worry. So he made his disclosures specific and ownership-forward.

He is specific about what the AI actually did versus what he did: not "AI wrote this," but "I had AI draft an initial version, then I reviewed it, checked the numbers, and adjusted the tone to sound like me." He emphasizes his review and customization, because that is what shows he owns the output rather than the machine. He explains why he used it, framing it as intentional efficiency, not laziness: "I used AI to get a fast first draft so I could spend my time on the analysis that matters." He is honest about limits: "AI helped me draft this, but I verified every figure against the ledger." And he avoids three traps. He does not hide it and hope, because it looks worse on discovery. He does not over-apologize, because "I'm sorry I used AI" treats a legitimate tool as something shameful. And he never claims the AI did more than it did, because "the AI independently analyzed this" overstates its reliability and would be a real problem in a finance context where someone might lean on that claim.

Worked example: a disclosure checklist and a before/after trust survey

Samuel wanted more than instinct, so he built a short transparency disclosure checklist and tested whether it actually moved trust. The checklist had five yes/no questions he runs before sharing any AI-assisted work:

• Affects others? Does this output touch someone else's work, evaluation, or decision? If yes, disclose.
• Would they want to know? Would the person be surprised or upset to learn AI was involved? If yes, disclose now.
• Did I name my role? Have I said specifically what the AI did and what I did?
• Did I protect data? Did I anonymize names and keep confidential figures out of the tool, and can I say so?
• Did I verify? Have I checked the facts and numbers myself, and am I comfortable owning the result?

To see whether transparency actually helped, Samuel ran a tiny anonymous pulse survey with his nine team members, one question, "I trust that my manager communicates honestly about how he works," scored 1 to 5. The baseline, taken the week after Priya's question, came back at an average of 3.1 out of 5, with two people at a 2. That was his "before."

Then he applied the checklist for six weeks. He held one open team meeting where he laid out exactly what he uses AI for and what he does not. He started tagging AI-assisted drafts with a one-line note on his role. When he used AI to find themes in collected team feedback, he told the team he had anonymized every response first, kept names out of the tool, and made the judgment calls himself. He stopped narrating trivial private use, so the disclosures he did make carried weight.

The follow-up pulse six weeks later averaged 4.4 out of 5, with nobody below a 4. A jump from 3.1 to 4.4 is a 1.3-point gain, roughly a 42 percent improvement on the baseline, and the two skeptics had moved to a 4. The number that mattered most to Samuel was qualitative: in the next round of one-on-ones, two people volunteered that they now felt comfortable using AI on their own reconciliations because they had seen him do it openly. The checklist did not just protect trust; it spread responsible use.

Putting it to work: real conversations

The checklist showed up in concrete moments. When Priya later asked again whether AI had touched a team message, Samuel answered cleanly: "Yes. I had AI generate a first version of what I wanted to say, then I reviewed it, rewrote it to sound like me, and made it specific to our situation. It is informed by AI but shaped by my thinking." Honest, specific, ownership clear.

When he used AI on the team's feedback data, he got ahead of the worry: "I used AI to help spot themes in the feedback we collected. I anonymized it first, so no names or identifiable details went into the tool. It helped me see patterns faster, but I read everything and decided what matters." Honest, and it surfaced the safeguard before anyone had to ask.

In his transparency meeting he set boundaries out loud: what he uses AI for (drafting, organizing, brainstorming), what he does not use it for (decisions about people, hiring, real feedback, though he may use it to organize his own thinking first), how he reviews everything, how he protects data, and a standing offer that if your work is involved he will tell you, and if you have concerns the door is open. And when his VP asked what he was doing with AI, he answered the same way: honest, referencing the company policy, naming his safeguards, and pointing to the hour a day it freed for higher-value analysis.

Two details made that team meeting land better than a simple announcement would have. Samuel named the organizational policy he was working inside, so his boundaries were not just personal preference but something anyone could go read for themselves, and he closed by explaining the point of the whole exercise: the time he saves goes back into developing his analysts and solving the harder problems, not into doing less work. Then he stopped talking and asked for questions and concerns, which is the part most managers skip. Inviting the objection is what turns an announcement into a conversation.

The judgment checkpoints behind it

Underneath the checklist sit a few questions Samuel keeps asking himself. Would they want to know? If yes, tell them. Would they be upset to find out later? If yes, disclose now. Is there anything about this use that would look suspicious if surfaced? If so, be extra transparent. What is the honest framing, the true story of who did what? And finally, am I comfortable with this becoming known? If the answer is no, that discomfort is information, and he reconsiders the approach rather than the disclosure. These are not abstractions; in finance, "would this look bad to an auditor" is a question with real teeth, and the same instinct serves him with his team.

Transparency, Samuel came to see, is not just etiquette. It is how he stays accountable for his own AI use, how he gives his team permission to ask questions and raise concerns, and how he models the responsible behavior he is about to ask them to adopt. The hour a day the tool saved him was never the real win. The real win was that his team could see exactly how he worked, and trust it.

Practice and reflection

Transparency is a habit, and habits form faster when you rehearse them against your own situation rather than in the abstract. Work through these before your next AI-assisted piece of work leaves your desk.

• Audit your recent use. Take three occasions in the past month where you used AI. For each one, should you have disclosed it, and to whom? What would you say now if someone asked?
• Write your framing. In your own words, in two or three sentences you would actually say out loud, how do you explain your AI use to your team?
• Set your boundaries. What will you use AI for, what will you refuse to use it for, and where exactly does your line on disclosure sit? Write it down; a boundary you have not articulated will not hold under pressure.
• Plan the conversation. Sketch the team meeting Samuel held. What are your five or six points, and what is the honest answer to the question you are most dreading?
• Anticipate the concerns. What is your team actually worried about with AI, whether or not they have said it? How would you address each worry without dismissing it?
• Interrogate the principle. How does transparency about AI use change trust, in your own experience? Where is the real line between simply not mentioning something and actively deceiving someone?

Where to take this next

Samuel turned the lesson into four standing commitments, and they are a reasonable place for you to start too. He wrote himself a personal AI policy, a single page covering what he uses AI for, what he does not, how he verifies output, and how he protects data, so his answers stay consistent instead of being improvised each time someone asks. He held the team conversation rather than waiting for another Priya moment. He kept practicing on real work rather than hypotheticals, because disclosure only gets comfortable through repetition. And he added a two-minute weekly reflection after his AI-assisted work: what worked, what did not, and what he would do differently next time.

That reflection habit is also how the earlier chapters of this level pay off. By the time you have foundational knowledge of what AI is, an eye for where it can genuinely help, practical skill with the tools themselves, and the responsible awareness this chapter builds, you have the whole of Level 1 in place, and you are ready to go deeper into specific techniques for managerial work.

Related lessons

• Mapping Your Daily Workflow gives you the framework for seeing which parts of your week AI could realistically touch. It is worth doing before your transparency conversation, because it tells you exactly what you will be disclosing.
• Your First AI Assisted Task is the practical companion to this lesson. Run a real task end to end using that workflow, then disclose it using the checklist here; the pairing is how the habit sticks.
• Level 2: AI-Assisted Use is where this goes next. Once your foundations, opportunity spotting, tool skills, and responsible awareness are in place, the next level deepens your technique with specific tools for managerial work.

Key Takeaways

• The discovery hurts more than the disclosure. If a colleague finding out later would feel like deception, that is your cue to say it now, openly and on your own terms.
• Disclose when it affects others, asks for their involvement, or goes public. Private, low-stakes use you reviewed yourself does not need a broadcast, but answer honestly if asked.
• Be specific, not vague. "I used AI for this" raises worry; "I drafted with AI, then reviewed, verified the numbers, and rewrote it to sound like me" shows you own the output.
• Use a short disclosure checklist. Five questions, does it affect others, would they want to know, did I name my role, did I protect data, did I verify, turn transparency from a gut call into a repeatable habit.
• Measure trust, do not assume it. A quick anonymous pulse before and after made the gain visible: Samuel's team moved from 3.1 to 4.4 out of 5 in six weeks of consistent, specific disclosure.
• Do not over-apologize or overstate. AI is a legitimate tool, so frame it as intentional efficiency, and never claim the AI did more than it did.
• Protect data and say so. Anonymize names and keep confidential figures out of the tool, then tell people you did; the safeguard is half the reassurance.
• Write your own AI policy and reflect weekly. A page on what you will and will not use AI for, plus two minutes of review after each use, keeps your answers consistent and your practice improving.
• Your transparency models theirs. When your team sees you use AI openly and responsibly, they feel safe doing the same on their own work.