Don’t get bot hurt. Get bot even.

This month, Bot Hurt is looking at the new rules of living with AI without making everything weird.

First up: Do you need to tell people when the bot helped?

Somewhere between spell-check and ghostwriter, the rules got fuzzy.

AI can clean up an email, tighten a resume, brainstorm a headline or turn a pile of notes into something coherent. It can also write the whole thing while the human takes a victory lap for pressing Enter.

Those are not the same. But somewhere along the way, “AI helped” became a catchall phrase for everything from deleting an unnecessary adjective to quietly outsourcing the entire assignment.

The etiquette is still buffering.

The disclosure dilemma

The obvious answer seems to be transparency. If AI helped, say so.

Except humans have managed to make that complicated, too.

A small January 2026 study posted on arXiv looked at how readers responded to AI disclosures in news stories. Researchers compared no disclosure, a simple one-line disclosure and a more detailed explanation of how the bot was used.

The simple disclosure did not reduce trust. The detailed version did.

Here’s the glitch: About two-thirds of the participants still preferred the detailed disclosure.

So, great. People want transparency. They just may trust you less when you give them too much of it.

Very helpful, humanity.

Not every comma needs a confession

The problem is not that people are using AI. It is that the phrase “AI-assisted” covers an enormous amount of territory.

There is a difference between asking a bot to catch a typo and asking it to write a recommendation letter for someone you barely remember supervising. There is a difference between organizing meeting notes and outsourcing the judgment that is supposed to come after the meeting.

Nobody wants a CVS-long receipt attached to every polished email. Drafted with assistance from one cup of coffee, Microsoft spell-check and a robot who removed three unnecessary adjectives.

That would get weird fast.

But when the bot materially shapes the work, the stakes change. A client, reader, hiring manager or coworker may reasonably want to know whether they are looking at your judgment, your expertise or a very confident remix.

A better test

The new etiquette does not require a full AI confession booth. It requires judgment.

If the bot cleaned up your sentence, carry on.

If it substantially wrote, analyzed or helped decide something another person is relying on, a heads-up is reasonable.

And if the work is supposed to reflect your effort, expertise or point of view, do not hand the bot the steering wheel and quietly climb into the trunk.

Final bot thought

The real question is not whether AI touched the work.

It is whether the human still owns it.

Bot Talk: Uber gets surge-priced by AI

Uber told employees to use AI as much as possible. Then the company got a familiar surprise: the meter had been running the whole time.

The ride-hailing company reportedly burned through its annual AI budget in four months after encouraging employees to experiment with tools such as Claude Code and Cursor, according to TechCrunch.

Uber even ranked internal usage on leaderboards, which is one way to turn tokenmaxxing into an office sport.

Now it has introduced a $1,500 monthly cap per employee for each agentic coding tool.

To be fair, this is not an argument against using AI. It is an argument against treating the bot like an all-you-can-eat buffet and acting surprised when the kitchen sends a bill.

Companies are entering the next phase of the AI boom: less “use the bot everywhere” and more “show us what the bot actually did.”

Efficiency still counts. But apparently, so do tokens.

Uber got a taste of surge pricing.

🚀 Coming up next week …

Sometimes the bot is working quietly in the background. Sometimes it steps forward and starts talking for the human.

Next week, Bot Hurt looks at the new courtesy gap: When does someone deserve to know they are dealing with AI?

Don’t get bot hurt. Get bot even.