Employees in China trying to use AI to get their co-workers fired so that their own jobs stay safe

In the age of AI everyone wants to save their own jobs, even if that means their co-workers losing their jobs. Or so seems the trend in some Chinese companies, where employees are increasingly training AI systems — but on the skills and competencies of their coworkers.

In Short

• Some Chinese employees training AI to do jobs of their co-workers
• They believe this will help them to keep their own jobs safe
• In response, some other employees have created anti-AI training manual

AI is making some jobs redundant. And as it does so, a number of Chinese employees are hoping that it would be doing so to the jobs their co-workers do. So that their own jobs can stay safe. To ensure that when a company thinks of job cuts due to AI, they are apparently training AI systems in their companies that can do the job of their co-workers. Bizarre and almost like a Black Mirror episode, but it seems to be a trend and now some of the AI training files that are part of the trend are viral on Github, a developer-focused platform.

If you’ve been following the growing conversation around AI and jobs on social media, you may have come across the term “skill distillation.” Put simply, it refers to breaking down how a person works, their processes, decisions, and methods, into clear, structured steps that an AI system can learn from and replicate. In effect, it’s about training AI to behave more like humans, with the potential to eventually take over parts of their roles.

In China, skill distillation is getting big. According to reports, many companies are asking employees to document their work in detail, from workflows and decision-making to internal communication. While this is being framed as a way to improve efficiency and knowledge sharing, the same data can also be used to train AI systems.

Employees, however, are beginning to recognise this pattern. The result? Some are now using the AI system to their own advantage and, in certain cases, against their co-workers.

A tool called “colleague.skill” is going viral on GitHub. It lets users analyse how an employee works by mapping their digital footprint, including chats, documents, and emails, and then turning that data into an AI agent that mimics their workflow.

According to reports, workers in China are using “colleague.skill” to record exactly how their co-workers do their jobs. They are turning their colleagues' daily tasks into simple digital guides that AI can follow so that those roles are easier to automate. The aim appears to be replacing co-workers with AI, while making themselves more valuable and secure in their own roles.

The counterattack too is happening

In response to "colleague.skill", a female developer in China has reportedly created a counter-tool called "anti-distillation.skill". In a viral video, the blogger demonstrates how the tool helps employees protect their expertise from being harvested by AI.

The tool essentially rewrites standard work documents of the employees so it remains clear and professional to a human reader, while blurring or omitting the most critical details. These are typically the parts AI systems rely on to learn how someone does their job.

As a result, the content appears complete on the surface but is less useful for training AI. In some cases, the tool adds vague or general language, making it harder for AI to understand patterns. It can also let users control how much information they want to hide, depending on how cautious they want to be.

- Ends