Employees Train AI to Replace Them, Then Sabotage It

In the ever-evolving landscape of Chinese tech companies, a fierce and covert corporate struggle is underway. Among various tools and strategies, artificial intelligence has emerged as a formidable instrument in the sphere of workplace dynamics.

As AI technology advances, many employees are taking matters into their own hands, attempting to make their coworkers redundant before official layoffs are announced. The approach is as straightforward as it is ruthless: analyze the tasks of a colleague, meticulously document their procedures, input this information into an AI system, and showcase the resulting AI agent to management as evidence of the job’s automability.

Managers are not left out of this drama. In numerous organizations, bosses are instructing employees to “document their knowledge” and compile detailed skill inventories. While this is marketed as a move towards team efficiency, the underlying objective is often to train AI replacements for their own team members.

This trend has escalated rapidly, fueling a competitive arms race on GitHub.

The “Colleague Skill” Offensive

A trending GitHub initiative known as Colleague Skill (colleague.skill) epitomizes this new battleground. This tool enables users to extract chat histories, emails, documents, and workflows from popular Chinese workplace apps like Feishu (Lark) and DingTalk. It subsequently consolidates this data into a reusable AI agent that mimics a specific colleague’s work habits, decision-making processes, and even personality traits (like their preferred emojis).

Initially intended partly as satire, this project gained immense traction. Employees realized they could use it to “clone” their coworkers and illustrate to management how easily their positions could be automated—hopefully at the expense of the colleague while safeguarding their own jobs.

The Counterstrike: Anti-Distillation Tools

Aware of the risks, employees quickly reacted. When required to describe their job responsibilities for “knowledge sharing,” a detailed account could potentially serve as the foundation for training their own successor.

In response, another GitHub project—the anti-distillation (or anti-distill) tool—rose to prominence almost instantly, serving as a means of defensive sabotage.

Here’s its operation: when a colleague or superior requests a breakdown of your tasks, you input your actual response into the tool. Instead of producing clear, actionable instructions, it transforms everything into **polished corporate jargon** that appears comprehensive to human readers but provides little substance for AI training.

Real example:

Original technical rule:
“Redis keys must have a TTL set. If there is no TTL, the Pull Request is immediately rejected.”

After the anti-distill tool:
“Usage of caching should comply with team standards.”

The tool features varying levels of intensity—light, medium, and heavy—depending on how closely your manager is monitoring. The outcome is documentation that appears satisfactory while safeguarding your actual expertise.

Also read:

A Symptom of Deeper Anxiety

This digital battleground is reflective of increasing anxiety within China’s tech landscape. In an atmosphere filled with fierce competition, constant layoffs, and swift AI implementation, many white-collar workers feel their skills have transformed from assets to liabilities—a potential threat that can be exploited against them.

What began as excitement surrounding AI productivity tools has devolved into a period of introspection. Employees who once eagerly engaged with tools like OpenClaw now find themselves entrapped in a dystopian cycle: tasked with building systems that may ultimately jeopardize their own jobs.

Discussions sparked by these GitHub tools have spread widely across Chinese social media, accumulating millions of views and likes. While some find dark humor in the situation, joking about “cyber-immortality” for laid-off colleagues, others caution that framing human capabilities as mere “skills” that can be replicated and discarded poses significant ethical and labor concerns.

As one developer observed, in the age of AI, the cardinal rule for survival in the Chinese tech workforce might soon be—never make your documentation too clear.

This is the new corporate arena. The tools are open-source, the strategies are cutthroat, and the ultimate goal is simple: to secure your job for another quarter.