I spent enough of the late nineties and early two-thousands in Photoshop to know what that work used to cost. Compositing of that quality was a half-day at minimum and a senior retoucher’s calendar. The output on my screen was indistinguishable from what a person like that would have delivered, and it had taken less time than reheating a coffee.

Then the client’s CEO came back with a note that got forwarded on to me. People attending the conference like golf, right? They’re the ones who are going to buy the service, so we should adjust the imagery to match them. Let’s change the setting to a golf course. Swap the families out for a couple of guys playing a round.

We were already a few days behind on getting the materials to the printer, so I just changed it. The new image was on the screen in another ten seconds. The families were gone. In their place: two men in pastel polos on a fairway, the product in the foreground, the light still matched, the work still acceptable. No half-day. No retoucher. No phone call to ask whether the swap was a good idea. Nobody on the team paused to consider what the image now said about who the product was for and who it wasn’t. The tool did not need to be talked into it. I did not need to be talked into it. It happened the way you take a sip of water.

I sat with the file open in front of me. I had done thirty years of work to be the person who would have asked the question. The question did not need to be asked, because the tool did not need to be persuaded. The thirty years was still sitting in the chair. The room did not need it.

Geoff Curtis, until recently Executive Vice President at Horizon Therapeutics, wrote a piece in Fortune in April. He had lost his role of nearly thirty years in an acquisition. The new owners did not need a Chief Communications Officer with the kind of long-form judgment Curtis had built. They needed less of him, and then none of him.

He gave the feeling a name: professional identity purgatory. Borrowed from the Catholic frame his readers might recognize: not heaven, not hell, a passage of purification. The space between losing what your career had made of you and not yet knowing what the next version of you is going to be.

I read it carefully. Curtis was describing the version of the condition where the door closes behind you and the badge gets deactivated. There is another version of it, harder to see because the person experiencing it is still going to work. Still on the calendar. Still getting paid. Still, by every metric anyone is measuring, doing the job. And quietly, behind their own ribs, no longer feeling like the job is theirs. That is the room this essay is trying to name. Not a quick moment, but a slow but steady weather system that settles in over time, and the people inside it cannot always tell when it arrived or how long it has been there.

It builds in the proposal that goes out on a Friday afternoon and gets the work without any pushback, when ten years ago the same proposal would have been a fistfight you sharpened your case for, and you cannot quite tell whether the room got easier or you got better or the bar got lower or something else entirely.

It builds in the client who praises a deliverable and means it, and the praise lands on the desk like an envelope of correct paperwork. You sign for it. You file it. You cannot feel the part of you that earned it.

It builds in the younger colleague who learned, in eighteen months, the thing that took you ten years to learn. They are not a fraud. They are good. The eighteen months is real. You watch them and you cannot decide whether the years of grinding through bad work were necessary or whether they were the price of being born early.

It builds in the meeting where someone asks how the deck got produced and you have to think about what to say. The honest answer is not that AI wrote it. The honest answer is that you used AI for the parts where you used to use junior people, and you would not have caught the difference yourself. You can describe the calibration you brought. You can describe the choices you made. You can describe the parts where your judgment ran the room. But the calibration moved fast enough that even you have stopped noticing it, and the moment it ran has already closed.

It builds in the client who does not know whether a tool drafted the language and increasingly does not seem to care. They are not being cynical. They are being efficient. They were always paying for the outcome. You believed you were being hired for the person who produced it. The math was never quite that, and the math is even less that now.

It builds at the end of the day. You used to leave the office with a particular kind of tiredness, the kind that comes from carrying something difficult and putting it down. Now you leave with a different kind of tiredness. It does not have a clear shape. You cannot tell whether it is the work or the absence of friction in the work or something underneath both, and you cannot tell whether you are supposed to be relieved about it or worried about it, and after a while you stop trying to name it and just go home.

None of this is grief about losing the job. The job is fine. The bookings are good. The reviews are warm. The slack is full of compliments. Curtis’s purgatory is a moment. The one I am trying to name is a weather system, and the people inside it cannot always tell whether the sky has changed or whether they have gotten worse at looking at it.

I have been in this weather before, in a different decade and a different city.

I lived in Istanbul in the middle of the twenty-tens, and on my evening commute I used to watch the city pull the day’s color out of itself, going from blue to pink to amber over the minarets. Orhan Pamuk has a word for the mood that settles into a city like that one: hüzün. Pamuk borrowed it. It describes a melancholy that is not private. It belongs to a place rather than a person. It comes from living among the remnants of something that used to be more than it is now, and from the fact that everyone around you can feel it too, and nobody quite talks about it.

I learned hüzün as a tourist’s word for Istanbul. It turned out to be portable.

The mood I felt on that commute had nothing, structurally, to do with what I am feeling now in a different chair in a different country. The decade is different. The trigger is different. The city is not even the same continent. The shape, though, is the same. A communal awareness, mostly unspoken, that the connective tissue between the people doing the work and the meaning that used to come back to them has gone slack. You can feel it in the room. You can hear it in the way people start sentences when they are talking honestly. You can see it in the way they stop.

Hüzün has come home. It is sitting on a thousand desks in a thousand offices on a Tuesday morning, and the people sitting at those desks have learned not to bring it up.

I do not think the feeling is grief that the work is going away. The work is going to be fine. There is going to be more of it, faster, and the people who can use the new tools well are going to do well.

I think the feeling is grief about the line between the work and the person who does it. For most of my career, the line was thick enough to lean on. A piece of work that came out of my hands had a particular grain to it. Clients knew it. I knew it. My team could spot the parts I had pushed against. The output came from a specific person, sitting at a specific desk, who had thought about the room and the client and the year of doing similar work that informed this one. The thinking was visible inside the output because the output was slow enough to carry it.

The tools have moved the work fast enough that the grain no longer travels with it. The thinking still happens. I can feel it, in the moment, when I am calibrating a prompt or rejecting a draft or making the small directional call that turns the next ninety seconds of generation into something usable. But the thinking happens at a speed that does not leave a mark, and the output that arrives does not show its work. By the time I have shipped, even I cannot quite find where I was in the thing.

That is the purgatory. It is not about losing the job. It is about losing the visible link between yourself and the job, while still doing the job, every day, on time.

What I built across thirty years was not, it turns out, the ability to produce the output. The output is increasingly available to people without the thirty years. What I built was the ability to feel my way through a room and bring back something that fit it specifically. That ability is still there. It is still being used. I might be the only person in the room who can see it being used. And on the days when I cannot feel it being used either, the chair I am sitting in starts to feel like it has nothing to do with me.

What is happening inside one person is a smaller version of what is happening across a whole field. The career is what becomes a person over time. The work is the receipt for the years. When the receipts stop carrying the weight, even though the work is still happening, the person and the career start to come apart, and you sit at your desk wondering which one is still you.

If you have been feeling some version of this, the name for it is professional identity purgatory. Curtis named it for the people whose door closed. I am borrowing it for the people whose door is still open, who are still walking through it on Monday morning, who are still, by every reasonable measure, fine.

The recognition is the whole point. There is nothing wrong with you for feeling this. The room you are in has a name now. It is not just yours, and it is not new, and it is not, in the end, a problem about AI.

It is a problem about what becomes visible when the speed of the work outruns the time it takes to be a person doing it. Sit in it.

The room is more crowded than it looks.

“Just as a heads up: I used Claude for the first pass and then cleaned it up from there.”

Nothing happens.

Then someone unmutes. Then someone else. The agenda moves on. But the room computed something in that silence, and everybody knows it.

That silence isn’t awkward uncertainty. It’s organized. And once you see the structure of it, you can’t unsee it.

Here’s what’s running simultaneously in a room when someone discloses AI use with no established norms around it: some people go quiet because they’re afraid — not for the person who said it, but for themselves. If disclosure becomes welcome here, it becomes expected. Others stay quiet because nobody else is affirming it, so it must not be what people do here. Some hold back out of consideration: validating the disclosure would put pressure on colleagues who haven’t made the same choice. And some stay quiet because silence is strategically safer than establishing a precedent either way.

That’s four separate calculations running in real time, in the same room, from the same silence. A policy about AI disclosure addresses zero of them.

The pressure context matters. Most of us are living inside a specific contradiction right now. The organization has told you to use AI to “10x your output.” Not using it signals you’re falling behind. At the same time, admitting you used it feels like a confession. Not a method note. Not a workflow detail. A confession.

There’s a Reddit thread that keeps circulating: someone’s boss now expects more deliverables per week because “we have AI now.” The thread has thousands of upvotes. It landed because it’s true. The mandate is to use the machine. The credibility cost of being seen using the machine is separate, real, and officially unaddressed.

This is the disclosure asymmetry: admitting AI use feels like vulnerability; not using it feels like failure. There’s no winning answer inside that frame.

So companies buy tools. They set up anonymous feedback channels, attribution systems, transparency dashboards. Remove the social cost of disclosure, and people will speak.

Mostly, they won’t. Not where it matters.

In small, high-accountability teams, the kind where disclosure actually means something, anonymity is implausible. Everyone knows who owns what project, who runs which deck, who wrote which doc. You can engineer something technically anonymous and the room still knows who it was. What changes the calculation isn’t anonymity. It’s something that has to exist before the standup starts.

The silence after “I used Claude for the first draft” isn’t a disclosure problem. It’s a trust architecture problem.

The disclosure moment doesn’t create the silence. It surfaces whatever the team’s psychological safety infrastructure can, or can’t, support. If the infrastructure was there before the standup, the admission lands as information, not as a verdict. If it wasn’t there, no policy changes what happens in the room.

Researchers call this “license to speak…” pre-authorized permission to raise concerns, make admissions, establish norms. It’s built by leadership behavior before any policy is written. Not by declaring that disclosure is safe. By demonstrating, through accumulated small moments, that honesty about uncertainty is welcome here.

The one mitigation that actually shifts team behavior isn’t a framing trick or a transparency tool. It’s what one longitudinal study of a software development team describes as “collective validity”: when enough people are using AI routinely and openly, disclosure shifts from confession to professional norm. Not through policy. Through precedent. Over time, through routine collective use, the question changes from should I say this? to how would I not mention it?

Someone has to go first. And how that first moment lands depends entirely on what was built before they opened their mouth.

I don’t have a clean answer for who that first person should be, or how to protect them in rooms where the scaffolding wasn’t built before the conversation arrived. The research doesn’t either. The honest position is that in most organizations right now, the infrastructure wasn’t built before AI disclosure became a daily question. The moment came too fast.

So the standup is happening anyway. In rooms with no established norms. With people running four different calculations in real time about whether the right move is to say something or say nothing.

The silence isn’t awkward. It’s a signal. And like most signals, it’s telling you something about what got built before the moment arrived, not about the moment itself.

What’s the actual norm on your team right now? Not the policy. The norm: the thing people do when nobody’s watching and nothing’s being written down.

It’s that moment I want to talk about. Not whether you’re allowed to use AI. You are. But whether you know what to do with the guilt that comes after.

Sixty-two percent of knowledge workers experience impostor syndrome, even before AI arrived. Now it’s amplified. A researcher at NYU named Adam Brown published a paper in March of this year that named what you might be feeling: “Authenticity Anxiety.” It’s a persistent preoccupation with a two-part question: Is this real? And is this *mine*?

The real question is: Is what I’m reading actually from a human? The internal part is yours: Is this work actually mine, or have I just run a prompt and dressed it up?

Here’s the thing, though. The fraud isn’t in the tool. It’s in the silence. You can use AI ruthlessly and remain credible. But only if you can say where you ended and the machine began. That’s not permission — that’s clarity.

Why Silence Costs More Than Honesty

You’re probably thinking: If I admit that I used AI, people will trust me less.

That’s the wrong question.

The real risk isn’t detection. People think they can detect AI content, but the research shows that they can’t. Seventy-six percent say it’s important to identify AI-generated material. Twelve percent can actually do it. That gap isn’t closing. It’s widening. Confidence is rising while accuracy declines. Your audience is worse at spotting AI now than it was two years ago, and they’re *more* confident about it.

So here’s the trap: they’ll probably never know you used AI. But if they find out later, the damage is different. A company called Bynder tested this. In a blind study, fifty-six percent of people preferred the AI-written article. It was better. They liked it. But the moment they learned it was AI-generated, fifty-two percent reported less engagement. Twenty percent called the brand untrustworthy.

The penalty wasn’t for using the tool. It was for the surprise.

Researchers at Northwestern ran thirteen separate experiments with over five thousand participants. They found that retroactive disclosure of AI involvement erodes trust by sixteen to twenty percent. But here’s the asymmetry: proactive transparency (telling people upfront) costs significantly less. The damage comes from the moment of discovery, not the tool itself.

This is the key insight. Your credibility isn’t fragile because you used a tool. It’s fragile because of the gap between what the audience assumes and what’s actually true. Close that gap deliberately, and you’ve moved from defensive to strategic.

Think about it. You’re not at risk of being exposed. You’re at risk of being discovered by surprise. And that surprise is what destroys credibility. The reader trusted the work because they trusted you. The moment they realize you withheld information about your process, that trust fractures. Not because you used AI. Because you didn’t say so.

There’s something elegant here: the antidote to the anxiety is the same as the business case. Transparency doesn’t destroy credibility. It builds it. And it costs less than the alternative.

Transparency is the upgrade, not the downgrade.

The Audit: Where’s Your Contribution?

If disclosure erodes less trust than secrecy, the question becomes: How do I know what to disclose?

Three questions. Answer them honestly, and you’ll know whether you’re a fraud.

Question One: What’s the original problem, insight, or experience that’s yours alone?

This is the foundation. What did you know or live through that no language model was trained on? I’ve spent thirty years watching clients navigate brand strategy. I’ve seen what sticks and what fades. I’ve made the weird conceptual connections between how people trust and how institutions fail them. I notice things. My brain makes odd links between concepts that AI simply can’t replicate.

That’s not modesty. That’s the bedrock. AI has no lived experience. It can synthesize, but it can’t originate what only you have seen.

Write down what’s yours. The specific insight. The years of work. The thing you know because you were there.

Question Two: Where did AI handle the grunt work?

I’m not shy about this. I use AI for research. I feed it policy papers and extract the useful signal. I ask it to summarize, to find patterns, to organize my thinking. There’s too much information and noise out there. More than any human could manually process. AI reads through stacks of material and surfaces what matters. It talks to me like a third-party observer and makes connections I’d miss alone.

The point: be specific. Not “I used AI to help.” Say what task. “I used Claude to synthesize twenty sources into a framework outline.” “I asked it to find contradictions across four studies.” Operational leverage. Not cheating.

Question Three: What would change if you removed the AI from the process?

This is the honest question. If the answer is nothing, you might have a problem. But if the answer is “I’d spend weeks researching instead of moving to strategy work” or “the deadline would slip and the project would become untenable,” you’re clear.

For me: without it, I’d still be researching. I’d never reach the creative thinking stage. I’d be lost in a sea of research, and frankly, I’d probably just give up. The gap between idea and execution would be so large that most projects wouldn’t happen.

Here’s what that sounds like when you say it:

*My insights come from thirty years of building brands and watching trust systems. I use AI to synthesize research and extract patterns from policy, which would take weeks or months manually. Everything I write, every strategic decision, every frame — that’s entirely mine. The machine handles the information overload so I can think.*

That’s not a fraud. That’s transparency.

What This Unlocks

The moment you answer those three questions, something shifts. You stop performing. The cognitive energy you spent hiding the tool goes away. The reader gets honesty. You get clarity.

Credibility compounds when you stop hiding the mechanism. Three years of consistent, transparent practice with AI becomes verifiable *you*. The problem everyone’s worried about (deepfakes, the “liars dividend” where any evidence can be called into question) has one answer: sustained practice. Show up with the same voice, the same insight, the same vulnerable specificity for years. That cannot be faked at scale.

There’s a case I keep coming back to. Yuval Halevi built a consultancy called Growtika. His entire practice is published proprietary knowledge. He writes about how he thinks. He shows his work. He doesn’t hide the AI tools he uses. And he’s thriving because the question isn’t “did he use AI?” The question is “does he know something I don’t?” And he does.

Your audience knows you use tools. They know you’re efficient. Transparency about how you’re efficient builds trust. The practitioners you respect — Ann Handley, Ethan Mollick - they’ve already stopped hiding. Silence is isolation. Transparency is belonging.

This is where your work shifts. You can mention the use of AI without apology. You can build systems around the tools, not despite them. Set up voice checks so the work still sounds like you. Build fact verification into the process so credibility doesn’t hinge on accident. Create approval frameworks so clients understand where the thinking happened. These aren’t defensive measures. They’re confidence markers.

You can charge premium rates because the conversation moved from “did they use AI?” to “did they actually think?” The market splits into two tiers: commodity content (where AI wins) and human-led thinking (where you win). You’re in the second category. Transparency is part of the credential.

Start Here

You’re not a fraud for using AI. You might be one if you can’t say where you end and it begins.

Answer the three questions. Write down your contribution. Keep that statement someplace you can find it. The next time you hit send on something you built with an AI, you’ll know exactly what to say.

That clarity is the permission.