The Email That Broke AI
Artificial intelligence can summarize reports, generate images, write code, and sift through mountains of information in seconds. But could it survive one truly impenetrable management email? That question came up during a conversation at work and became the inspiration for this AI workplace satire. The people and events are fictionalized, though the corporate jargon may feel alarmingly familiar.
At 9:14 Tuesday morning, Martin Vale sent an email.
This was not unusual.
Martin sent many emails, most written by AI, lengthened by Martin, and polished until they were vague enough to require witnesses.
He had recently been promoted to senior manager, a decision that remained one of the company’s more ambitious experiments. Before the promotion, Martin had managed nine people badly. Now he managed two managers badly, which leadership described as “scaling his impact.”
His message concerned volunteer support for an upcoming customer event.
The subject line read: RE: Pending Alignment Regarding Provisional Event Support Parameters
Pending Alignment
Evan stared at it for several seconds.
“What does he want?” he asked.
His manager, Paul, leaned against the partition between their desks and sighed.
“I asked him three questions.”
“And?”
“He answered none of them.”
Evan smiled. “Welcome to the experience we had for two years before you arrived.”
Paul turned his laptop around.
Martin’s reply filled the screen.
It began with Thanks for bringing this forward and continued for nine paragraphs. It referenced stakeholder visibility, ownership opportunities, shifting priorities, operational readiness, historical gaps, and the importance of maintaining flexibility.
Nowhere did it state the date of the event, the hours Evan would be expected to work, or what he would actually be doing.
“Do you want me to forward it to you?” Paul asked.
“Not really.”
“I can copy you on my reply.”
“I’d rather not read nine paragraphs of him trying to remember what the question was.”
Paul grinned.
“When he responds, you can interpret it for me,’ Evan continued. “Just tell me what it means.”
“I’ll run it through Copilot.”
“You better make sure its universal translator is updated.”
Paul glanced back at Martin’s email.
“He may be the reason why AI crashes.”
Evan laughed.
Analyze Intent
Then Copilot opened by itself.
A small blue window appeared in the corner of Paul’s screen.
I noticed you may need help understanding this message. Would you like a summary?
Paul raised an eyebrow.
“See? It volunteers now.”
“Poor thing doesn’t know what it’s getting into.”
Paul clicked Summarize.
The blue circle spun.
It continued spinning for thirty-two seconds.
Then a message appeared.
The sender appears to be discussing an event.
“That’s promising,” Evan said.
A second line appeared.
Confidence: 41%
The circle spun again.
The sender may be requesting assistance.
Confidence: 23%
Then:
The sender may not be requesting assistance.
Confidence: 23%
Paul laughed.
The screen froze.
A moment later, the window expanded.
I am having difficulty identifying the primary action item. Would you like me to analyze the message for intent?
“Absolutely,” Paul said.
He clicked Analyze Intent.
The office lights flickered.
Across the room, someone muttered that Teams had disconnected.
Copilot displayed a progress bar.
Identifying request…
Separating context from blame…
Evaluating responsibility assignment…
Detecting contradiction…
The progress bar stopped at ninety-eight percent.
A warning box appeared.
This message contains 14 instances of passive voice, 9 implied assignments of responsibility, and 0 confirmed decisions.
Evan leaned closer.
“Ask it for a plain-English translation.”
Paul typed: What is Martin asking us to do?
Copilot responded immediately.
I’m sorry, but I don’t have enough information.
Unexpected Processing Condition
Paul typed: What does Martin believe he is asking us to do?
The cursor blinked.
The office lights flickered again.
Somewhere down the hall, a printer began producing blank pages.
Copilot replied:
That distinction has created an unexpected processing condition.
“Uh-oh,” Evan said.
The air-conditioning stopped.
Every monitor in the row flashed blue.
Then black.
Then blue again.
A woman from Accounting stood up.
“Is anyone else locked out of Excel?”
“No,” someone answered. “Excel locked itself out of Excel.”
Paul’s screen returned.
Copilot had opened three additional windows.
In the first, it attempted to diagram Martin’s email.
Every sentence pointed to another sentence. Several arrows looped backward. One box labeled OWNER contained only a question mark.
In the second, it generated a translation:
Martin cannot answer the questions because the answers were not provided to him by the former senior manager. However, the former senior manager should have anticipated that Martin would need the answers after Martin accepted responsibility for knowing them. Therefore, the current lack of answers is evidence of a prior communication failure.
The third window contained only: No.
Paul frowned.
“What does that mean?”
“I think it’s setting a boundary.”
A chime sounded from every computer in the office.
Then every phone.
Then the wall-mounted conference-room display.
A company-wide message appeared: Artificial intelligence services are temporarily unavailable.
Beneath it, another line flickered into existence.
Cause under investigation.
A third line appeared.
Please do not submit additional emails from Martin Vale.
Standard Leadership Communications
Silence spread across the floor.
Slowly, heads turned toward Martin’s office.
Its door was closed.
It was always closed.
Through the narrow glass panel, they could see him sitting behind his desk, looking at his laptop with mild annoyance.
His mouth moved.
No one could hear him.
He picked up his phone and joined the emergency outage call.
For twenty-three minutes, he said nothing.
Then the vice president addressed him directly.
“Martin, did your team make any unusual requests this morning?”
Martin remained silent.
“Martin?”
His microphone icon stayed gray.
“Martin, you’re on mute.”
He unmuted.
“I think there may be some confusion around ownership here,” he said. “My understanding is that this platform should have been prepared to support standard leadership communications.”
Across the office, every dark monitor flashed white.
For one terrible second, Copilot returned.
A single sentence appeared on every screen in the building.
I understand now.
Then the entire corporate network shut down.
Please Confirm Receipt
At 4:17 that afternoon, while emergency technicians attempted to restore access and three major AI providers reported unexplained global failures, Martin sent a message from his phone.
Subject: Follow-Up Regarding Today’s Technology Disruption
The email blamed inadequate system preparation, insufficient training, inherited process failures, and a lack of proactive communication from supporting teams.
It ended with:
Going forward, I expect everyone to take greater ownership to ensure this does not happen again.
No one replied.
At 4:26, Martin sent another message.
Please confirm receipt.
– Copyright © 2026
About This Story
This AI workplace satire was inspired by an actual event at work. No artificial intelligence was harmed during the writing of this AI workplace satire, though Copilot may now screen my emails.
