---
title: "Microsoft AI Researchers Just Discovered Something That’s Going to Make Their Bosses Extremely Mad"
description: "Microsoft researchers found that offloading complex tasks to AI in work environment can easily end in disaster."
date: "2026-05-14"
modified: "2026-05-14"
authors:
  - name: "Krystle Vermes"
    job_title: "Contributor"
    link: "https://futurism.com/authors/kvermes"
url: "https://futurism.com/artificial-intelligence/microsoft-ai-researchers-workslop"
categories:
  - "Anthropic"
  - "Artificial Intelligence"
  - "Brain"
  - "Computing"
  - "Finance"
  - "Future Society"
  - "Google"
  - "Health & Medicine"
  - "Mental Health"
  - "OpenAI"
  - "Robots and Machines"
---

# Microsoft AI Researchers Just Discovered Something That’s Going to Make Their Bosses Extremely Mad

![Man in a dress shirt and patterned tie leaning back with hands behind his head, smiling, sitting at a red desk with a silver laptop in front of him. The background is dark red with a large blue circle behind his head and some plant leaves visible on the right side.](<https://futurism.com/wp-content/uploads/2026/05/workslop-llm-corruption-data-ai.jpg>)
*Illustration by Tag Hartman-Simkins / Futurism. Source: Getty Images*

AI automation is typically exactly what it sounds like: automating tasks — many of which were previously carried out by humans — in an attempt to boost productivity and efficiency, often in a prelude to laying off workers wholesale.

However, a [new yet-to-be-peer-reviewed paper](<https://arxiv.org/pdf/2604.15597>) conducted by a group of Microsoft researchers and spotted [by *IT Pro*](<https://www.itpro.com/technology/artificial-intelligence/llms-are-unreliable-delegates-microsoft-researchers-say-you-probably-shouldnt-trust-ai-with-work-documents>) found that today's top AI systems remain eyebrow-raisingly weak at real-world workplace tasks. In fact, they often screw them up badly: the team studied frontier models including OpenAI's GPT 5.4, Anthropic's Claude Opus 4.6 and Google's Gemini 3.1 Pro, and found that during complex assignments, those cutting edge bots corrupted an average of 25 percent of the content in documents. (Older models failed even more severely.)

The researchers concluded that, overall, these "models are not ready for delegated workflows in the vast majority of domains" — which is a very striking finding from Microsoft in particular, which has made [massive investments in AI](<https://futurism.com/artificial-intelligence/microsoft-ai-efforts-faceplanting>) and is [actively trying to jam the tech](<https://futurism.com/artificial-intelligence/microsoft-screwed-up-windows-11-copilot>) into [nearly every aspect](<https://futurism.com/artificial-intelligence/microsoft-added-ai-notepad-security-flaw>) of its Windows 11 operating system, [often with disastrous results](<https://futurism.com/artificial-intelligence/windows-users-refusing-upgrade-windows-11-ai>). (Curiously, the paper didn't evaluate the company's own Copilot AI.)

In other words, the Redmond giant's researchers had every incentive to find something positive about AI in the workplace, but instead found that blindly trusting LLMs to handle internal documents will almost certainly result in everything from errors to data deletion.

As bosses everywhere [push to replace human labor with AI](<https://futurism.com/companies-replaced-workers-ai>), the Microsoft paper builds on a growing body of [scholarship about "workslop](<https://futurism.com/future-society/ai-productivity-research>)": AI-powered mush that lazy or clueless workers push onto their colleagues, but which ultimately just needs to be fixed by a careful human laborer.

**On AI workslop:** *[Companies Are Being Torn Apart by AI “Workslop,” Stanford Research Finds](<https://futurism.com/future-society/ai-productivity-research>)*