Before You Send That AI-Written Email: A Five-Step Credibility Check for Leaders
By Stacey Hanke, CSP® and CPAE Speaker Hall of Fame® inductee
Generative AI can help you write an email in seconds.
It can also make you sound like someone your employees have never met.
The problem isn’t that you used AI. The problem is that you allowed AI to speak for you.
AI can help you organize your thoughts, tighten your message and get past the dreaded blank page. What it can’t do is understand the full weight of your relationships. It can’t anticipate every employee concern. It definitely can’t take responsibility for the words sent under your name.
That part still belongs to you.
How AI-Assisted Emails Can Undermine Your Credibility
Your employees know how you communicate.
They recognize your vocabulary. They know how direct you are. They know whether you tend to get straight to the point or take the scenic route.
What happens when a leader who normally speaks plainly suddenly sends an email filled with phrases like “strategic realignment,” “organizational synergies,” and “future-state optimization”?
People notice the disconnect.
Even if your message is polished, it may leave employees wondering:
Did my leader actually write this?
Does my leader understand how this decision affects us?
Is all this vague language hiding something?
Can I trust the promises being made?
This doesn’t mean every polished email damages trust. It also doesn’t mean employees can always tell when you used AI.
The real risk is inconsistency.
That concern is supported by workplace communication research. In a study of 1,100 professionals, employees evaluated managers differently based on how extensively they relied on AI to write workplace messages. The findings suggest that AI assistance may improve polish while extensive reliance can affect perceptions of sincerity, care and trustworthiness.
When your written message sounds nothing like the person employees know, it creates distance. As Stacey explains in “Is AI Fracturing Your Credibility?,” the risk becomes even greater when your polished written communication doesn’t match how you show up in person.
AI-Assisted Communication Still Requires Human Accountability
Using AI doesn’t transfer responsibility away from you.
The National Institute of Standards and Technology’s Generative Artificial Intelligence Profile recommends defining responsibilities for human oversight and verifying sources and citations in generative AI outputs.
The framework applies to far more than executive emails, yet the principle is the same: someone still needs to own the output.
An AI-assisted message is communication developed with help from a generative AI writing tool. You might use AI to brainstorm, organize your ideas or clean up a draft. You’re still responsible for the final message.
AI can help draft the words.
It can’t own the consequences.
Use the CLEAR Check Before You Press Send
The CLEAR Check is a five-part review designed to uncover the problems a polished AI draft can hide: missing context, borrowed language, unverified claims, overlooked audience concerns and commitments you’re not prepared to keep.
Clear: Have you clearly explained what people need to know?
AI can organize the information you give it.
It can’t identify important context you never entered.
Before you send the message, ask:
Have I explained what’s changing?
Have I explained why it’s changing?
Have I said who will be affected?
Have I identified what happens next?
Have I acknowledged what I still don’t know?
A message can be completely accurate and still feel evasive. Accuracy won’t help if you leave out the information your audience needs most.
Language: Does this sound like you?
Read the message aloud.
Would you actually use these words in a meeting? Could you say them naturally in a conversation? Does the message sound like the same person employees hear Monday to Monday®?
Remove the inflated language. Cut the unnecessary formality. Get rid of anything you’d never say face to face.
The goal isn’t to make every message casual. The goal is to make it sound like you.
Evidence: Have you verified every claim?
Check every number, date, quotation, attribution and factual statement.
Click the links. Confirm they go to the right sources. Make sure those sources actually support what the message says.
Never assume that a confident sentence is an accurate sentence.
Does the message include a promise? Make sure you and your organization can keep it.
Audience: Does the message address what people care about and why?
Leaders often communicate from the organization’s point of view.
Employees read from their own.
They’re asking:
How does this affect me?
What do I need to do?
When will this happen?
Who made this decision?
Where can I ask questions?
What aren’t you telling me?
You don’t need to answer every possible question in one email. You should answer the questions you can reasonably anticipate. If you don’t have all the answers yet, say that. Tell people when they can expect to hear more.
Vagueness doesn’t make uncertainty disappear. It only makes people more suspicious of it.
Reputation: Will you stand behind every word?
Your name is attached to the message.
That makes it yours, even if AI wrote the first draft.
Before you send it, ask:
Can I defend this wording?
Can I explain this decision?
Can I verify these claims?
Can I keep these commitments?
Can I respond when someone challenges me in person?
If the answer is no, the message isn’t ready.
What the CLEAR Check Looks Like in Practice
Imagine your company is preparing to announce a restructuring.
An AI-generated first draft might say:
To position the organization for sustainable growth, we are implementing a strategic realignment that will optimize resources, improve operational efficiency and create future opportunities across the enterprise.
It sounds professional.
It also says almost nothing.
Employees still want to know whether jobs are being eliminated, which teams are affected, when the changes begin and where they can get real answers.
Here’s what a leader-reviewed version might sound like:
Beginning October 1, we are combining our customer support and account services teams. Twelve positions will be eliminated as part of this change. Employees whose roles are affected will be contacted individually today. No additional position reductions are planned as part of this restructuring. We will hold a companywide meeting at 2 p.m. to explain why we made this decision, what changes next and where employees can ask questions.
The second version isn’t better because it sounds more human.
It’s better because it gives people context. It answers the questions they’re already asking. It makes commitments the leader should be prepared to defend.
The exact language will depend on the facts. The standard shouldn’t change:
Clarity before polish.
When Leaders Shouldn’t Delegate the First Draft
Some messages need your judgment from the very first sentence.
Be especially cautious about turning the initial framing over to AI when you’re communicating about:
Layoffs or restructuring
Workplace injuries or deaths
Public apologies
Ethical or legal violations
Discrimination or harassment allegations
Significant organizational failures
Decisions affecting employees’ careers or compensation
AI may help you identify missing questions or edit a later draft. It shouldn’t determine the emotional posture, level of accountability or core message in a high-stakes situation.
The more consequential the message, the more visible your judgment needs to be.
AI Should Accelerate Your Thinking, Not Replace It
Generative AI is great for overcoming a blank page. It can organize your ideas, tighten your sentences and help you identify information gaps. You can even ask it to generate the questions an employee might have after reading your message.
That’s assistance.
It’s not ownership.
People don’t judge the tool you used. They judge the person whose name is attached.
Before your next AI-assisted message goes out, read it aloud and apply the CLEAR Check:
Clear: Have you clearly explained what people need to know?
Language: Does this sound like you?
Evidence: Did I verify every claim and commitment?
Audience: Does the message address what people care about and why?
Reputation: Will you stand behind every word?
Ask yourself one final question:
If an employee challenged this message in person, could I defend every word?
If the answer is no, don’t press send.
Pull up one recent AI-assisted message and run it through all five parts of the CLEAR Check. The gaps will show you where AI is supporting your communication, and where it may be speaking for you.
