Don't blame the model for your layoffs

Another week, another CEO blog post about how AI is the reason a few hundred people no longer have jobs. The framing is everywhere now: “AI made us more efficient,” “we're reshaping the org around AI,” “these roles are being automated.” It's dishonest, it's disrespectful, and it's a tell that the leader doesn't actually understand the tool they're hiding behind.

AI didn't fire anyone. You did.

The grammatical trick in every one of these announcements is the same: AI becomes the subject of the sentence. “AI is replacing roles.” “AI allowed us to reduce headcount.” The model didn't walk into a room and decide a P&L target. A human did. Naming the model instead of the decision launders a hard call into an inevitability.

The honest version is two sentences. We over-hired between 2021 and 2023. We're correcting, and we're investing in tooling so the people who stay are more effective. That's a leadership statement. The AI framing is a press release.

It's disrespectful to the people leaving

“Your role was automated” tells a person their seven years of judgment, relationships, and context were a script a model could run. They weren't. They were doing work the company chose not to keep paying for. Those are different sentences and they land differently on a resume, a LinkedIn post, and a kitchen table conversation that night.

It's worse for the people staying

The cost most leaders miss: every survivor of an “AI layoff” now spends part of every week quietly auditing themselves. “Is my function next?” “Should I be using Cursor more visibly?” “Is the new tool actually a timer on my role?” That tax is invisible on the spreadsheet and enormous in the work. Trust is the highest-leverage asset on any team and you just spent a quarter of it for a headline.

The cruel irony: shipping great AI products requires exactly the conditions you just destroyed. Experimentation, a willingness to surface where the model is bad, honest feedback on what to keep doing manually. A scared team doesn't do any of that. They hide.

What good leaders do instead

• Own the decision in plain language. “We are reducing headcount because revenue per employee is below where it needs to be. AI is part of how we plan to grow productivity from here, not the reason for the cut.”
• Separate the two conversations. Layoffs are a business call. AI adoption is a capability call. Conflating them teaches the org that the new tool is a threat, not a lever.
• Invest in the people who stay. Real time, real training, real permission to try the tools and report back honestly on what works. That's how you actually compound the productivity gain you announced.

The product lens

I run a B2B portfolio. Most of what my team ships in the next two years will involve AI somewhere in the stack. None of it gets better because the team is afraid. It gets better because they trust the company enough to push the model into uncomfortable places, find where it breaks, and tell me. Morale isn't a soft variable here. It's the eval set you ship against.

If you're a leader writing one of these posts, take the AI out of the sentence. If the decision still reads as defensible without the model in the subject line, send it. If it doesn't, you have a different problem than the one you're announcing.