I’m noticing something interesting with the professionals I work with. There are ongoing anxieties about using AI, but those anxieties are shifting.

The anxiety is no longer only about whether we should use AI. It’s increasingly about whether we’re using it correctly, acceptably, transparently, legally, ethically, and according to rules that are still being written.

You’ve heard me talk about professional and personal double binds before. This is a new one I’m hearing about every week, and it is big.

Organizations say: Adopt AI. It will increase productivity and help us stay ahead of the competition. It’s an experiment that has already (ironically) been proven by the very AI we are using. Don’t be afraid to experiment and think outside the box.

Oh, and btw, become AI fluent…quickly.

While simultaneously saying: Don’t put that information in there! Oh, don’t use that AI model; it hasn’t been approved. Don’t let AI write that. Make sure it is an original thought that comes from you and sounds like you. Disclose when you’ve used it, including why, when and how. But don’t disclose it in a way that undermines trust or negatively impacts how the company is perceived.

And whatever you do, don’t violate the policy we’re still developing. And what my clients come away hearing is: “Use it. But not like that!”

· A physician wonders whether using AI to summarize something makes her more efficient and cutting-edge, or totally irresponsible.

· An executive quietly uses it to think through a difficult decision but doesn’t mention it because he doesn’t know whether that will be perceived as innovative or lazy.

· An employee discovers a workflow that saves six hours and then wonders whether telling his manager will earn him recognition or scrutiny.

· A writer begins second-guessing perfectly ordinary sentences because someone might decide they “sound AI-generated” because they’ve written in groups of three or used em dashes.

And what I’m noticing developing as a result of all this is AI hypervigilance.

People are monitoring not only what they’re doing, but how their use might be perceived. They attempt to hide usage, over-explain it, or under-report it. They run their own writing through AI detectors. They ask AI to make something “sound less AI” and then become anxious that even that will be detected.

People maintain parallel workflows to prove what they did independently, which takes more time and creates more confusion. And increasingly, they wonder whether their own work still looks like their own work.

Does this sound like me? Does it look like me? Will other people believe it’s mine?

All questions that undermine confidence, promote imposter syndrome and negatively impact meaning. (Yes, I used a grouping of three, and it’s my own thought!)

The technology designed to increase efficiency can suddenly produce an entirely new layer of cognitive and emotional labor. This creates intellectual, emotional, somatic, and relational consequences. A tool designed to simplify our work becomes embedded inside a double bind that can actually complicate it.

And there is a fascinating organizational paradox underneath all of this:

Leadership is being told to adopt AI aggressively, so adoption is moving at the speed of competition. Meanwhile, governance is moving at the speed of committees. Human beings are being asked to live in the gap. The gap between speed and compliance. Between efficiency and perception. Between experimentation and permission. Between what the technology can do and what your organization has decided you’re currently allowed to do with it.

Here’s the bottom line:

We’re being told to move faster with a technology whose rules are still being written.

So what actually helps? Probably not another policy memo, she wrote tongue in cheek.

You have enough of those, and sometimes they become part of the problem. Each new one arrives sounding definitive and then becomes a placeholder for the next one. A better starting point is naming the double bind out loud, to yourself and to your team, instead of quietly absorbing it as evidence that you’re doing something wrong.

“We are being asked to move quickly while staying compliant with rules that are still evolving. Those expectations will sometimes conflict. I’m someone you can talk to about this.”

Naming the contradiction matters because once we recognize a double bind, we can stop trying to solve it entirely inside ourselves. Naming the truth is always powerful. It helps free us from the distortions, accommodations and performing that use us up and burn us out.

There is another distinction I think is especially useful.

We have been merging two very different questions:

“Is this a good use of the tool?”

and

“Is this use of the tool currently sanctioned?”

Those are not the same question. The first question requires autonomous judgment. Is the information appropriate? Is it accurate? Does using AI improve the work? Am I still thinking critically? Am I protecting confidential information? Am I accountable for the result?

The second question requires compliance. What does my organization currently permit? What needs to be disclosed? Which platforms are approved? What does the policy say today, and in what direction is that policy moving?

You can have excellent judgment about the first question while the second is still being sorted out by people whose job it is to figure it out. And sometimes good judgment will tell you not to use AI even when policy permits it. Sometimes good judgment will tell you AI would be exceptionally useful in a situation where organizational policy hasn’t caught up yet.

That tension is the bind.

Trying to eliminate it by becoming more careful, more vigilant, more controlling and more conscientious won’t necessarily make you safer. At some point vigilance becomes hypervigilance, and hypervigilance creates its own problems: hesitation, concealment, duplicated work, second-guessing, and exhaustion.

We need vigilance right now. We also need discernment…and accountability. But we do not need to turn every interaction with AI into a referendum on our professionalism, intelligence, integrity, or worth. The rules will continue to evolve. Organizations will learn. Policies will change. Some experiments will become standard practice, and others will prove to have been terrible ideas.

We are standing inside an unfinished structure.

Our work isn’t to pretend the structure is finished. Pretending certainty exists when everyone can see that it doesn’t is simply a modern version of The Emperor’s New Clothes. Our work is to develop enough judgment to work responsibly inside the structure while it is still being built. To tolerate some uncertainty without converting it into hypervigilance. To comply without surrendering our capacity to think.

And to use a profoundly powerful new tool without forcing ourselves into a double bind every time we do.