If you use AI and can't explain how you got there, is the work even yours?
There is a particular quiet you feel when you hand in good work you cannot quite explain.
The slide deck is sharp, the email lands, the summary is clean, and yet if your boss leaned over and said "walk me through how you got here," you are not sure you could. That feeling is authorship anxiety, the unease that the thinking might not have been yours even when the output has your name on it, and it is a different nerve from worrying you are a fraud.
The good news is there is one habit that reliably keeps the work feeling like yours, and a recent study points straight at it.
Here, let me show you what it is and how to build it into how you work.
Is this imposter syndrome, or something else?
It is something else, and naming it correctly matters because the two need different medicine.
Imposter syndrome is the fear that you are not good enough and will be found out. Authorship anxiety is quieter and more specific: you know you are capable, but you are no longer sure the work in front of you reflects your own reasoning, because a tool did a chunk of the thinking and you cannot retrace the steps. You can feel completely confident in your abilities and still feel this.
That is the tell.
A consultant I would describe as one of the sharpest thinkers I know told me she had stopped using AI for her client notes, not because the notes were worse, but because she could not reconstruct how she had arrived at them, and that bothered her more than she expected.
She was not worried she was a fraud. She was worried the work had drifted away from her.
This distinction is why the usual imposter-syndrome advice misses. "You earned your seat, you belong here" does nothing for a person whose actual question is whether they could explain their own reasoning out loud if asked.
What does the research actually say about AI and ownership?
The clearest signal comes from a study published in the journal Technology, Mind, and Behavior (Baldeo, April 2026), which surveyed 1,923 adults across the United States and Canada about how using AI affected their sense of ownership over their work. In that study, 58% of participants agreed that AI did most of the thinking when they used it.
Worth being precise here: that 58% is what people reported about their own experience, not a measurement of how much thinking AI actually does across the population. The finding that matters most, though, is not the worry. It is the protection.
The participants who actively challenged, modified, or rejected the AI's suggestions reported a stronger sense of authorship and greater confidence in the final work than those who took the output as given.
The act of pushing back was the thing that kept the work feeling like theirs.
Read that again, because it quietly rewrites the whole problem.
The risk to your sense of ownership is not that you used AI. It is that you accepted what it gave you without putting your own judgement through it. And your judgement, the thing you have spent twenty years sharpening, is precisely the part the tool cannot supply.
So how do I keep the work feeling like mine?
You build a small ritual of challenging the output before you accept it, and you do it every time, so that your reasoning is back in the loop by the time the work has your name on it. The study suggests the protection comes from the pushing back, so the move is to make pushing back a habit rather than an afterthought.
Here is the ritual I use and teach.
Call it the challenge-the-output pass.
Find one thing you disagree with. Read what AI gave you and locate a single claim, phrase, or choice that does not sit right. There is almost always one. If you genuinely cannot find anything to question, that is a signal you have not read it closely enough, not that it is perfect.
Ask it why, then push. Make it defend the choice, then tell it where you think it is wrong and watch whether it folds or holds. You are not being difficult, you are testing the reasoning, which is exactly what you would do with a junior colleague's draft.
Change at least one thing on your own judgment. Rewrite a line, cut a section, reorder the argument the way your experience tells you it should run. This is the step that re-enters your fingerprints onto the work.
Say the reasoning out loud. Before you send it, explain to yourself in one sentence why the final version is shaped the way it is. If you can do that, you can walk anyone through it, and the authorship question answers itself.
Done properly, the whole thing takes a few minutes, and it is the difference between work AI handed you and work you made using AI.
What if I genuinely can't reconstruct my reasoning?
Then you have found the exact spot where the work slipped away from you, and the fix is to go back one step rather than to feel bad.
Authorship is not about memorising every keystroke.
It is about being able to stand behind the decisions, and you can only stand behind decisions you actually made.
So when you hit a piece of work you cannot retrace, treat it as information, not as a verdict on you. It means the tool got further ahead than your judgement did on that task.
Take that one back, run the challenge-the-output pass on it, change what your instinct says to change, and notice how quickly the work feels like yours again. The reasoning was never gone, it was just waiting for you to put it back in.
That consultant, by the way, did not stop using AI for her notes. She started arguing with it first, and told me the notes got better and felt like hers again in the same week.
The fix was not less AI. It was more of her.
So the next time the tool hands you something good, before you accept it, what is the one thing you are going to push back on?