AI Clones of Departed Employees Spark Workplace Ethics Crisis in China

A former video editor named Li Yao discovered nearly a year after resigning that her cloned voice was still being used in her former company's advertisements and internal content—a troubling example of a broader workplace phenomenon where companies use artificial intelligence to capture and replicate employees' skills, decision-making logic, and personality traits into permanent digital assets. The practice, rooted in the concept of "model distillation," converts individual expertise into what companies term "digital employees," allowing firms to maintain worker productivity even after employment ends. Li learned of the voice cloning when former colleagues noticed her distinctive voice in ads played during internal review meetings, despite her having never consented to such use.

The trend reflects a dramatic shift in AI disruption within the workplace. Three years ago, artificial intelligence primarily threatened basic knowledge work and creative fields; today it targets individual work processes, judgment, and personality itself. Companies increasingly expect workers to translate their intuition, experience, and operational habits into prompts and workflows that can train AI systems—making the training of one's own AI replacement an implicit job requirement. Stanford intern Xu Kejia experienced this firsthand while writing animation scripts for a Chinese tech giant; her primary task was teaching AI to write with human-like quality, often at a pace slower than her own work. At Meta, the company collected employee keystrokes and mouse clicks for similar purposes, triggering backlash labeled an "employee data extraction factory," particularly during mass layoffs. Even executives are not exempt; digital twins are being developed to replicate high-performing CEOs' decision-making.

The corporate push has sparked a fierce counteroffensive among tech-savvy workers who developed open-source tools to subvert data extraction and reassert control over their digital legacies. Within a week of "colleague.skill"—an open-source AI agent designed to ingest a worker's digital footprint and mimic their operating habits—going viral, "anti-distillation.skill" emerged, allowing employees to "wash" knowledge documents by replacing core insights with "correct but useless nonsense" while preserving genuine expertise privately. The tool garnered over 4 million views and was described as "fighting magic with magic." Another tool, "keep-a-hand.skill," helps workers identify and protect their hardest-to-replace skills, such as critical judgment, while surrendering only standardized procedures. These resistance tools highlight the existential threat many workers face: being forced to hand over personal data and knowledge to train AI systems that may eventually replace them entirely.

The legal landscape remains murky and challenging for workers defending their digital identities. Chinese courts have begun ruling that AI-driven upgrades do not automatically justify unilateral employment changes, but defending against digital cloning remains legally complex and costly. Employment contracts typically do not explicitly address whether companies can use extracted labor data or personality traits indefinitely for AI training, creating a significant legal gap around property rights for such digital assets. Litigation is lengthy and expensive, with successful outcomes rare outside celebrity cases, and compensation often falls far short of actual damages. Li Yao, for instance, now spends evenings scrolling through videos to find and report her cloned voice—a grinding process as deleted ads are quickly replaced with new content featuring the same synthetic voice.

The debate reveals fundamental tensions between corporate efficiency and worker autonomy in the AI era. Under China's Civil Code, AI-generated voices or styles recognizable as belonging to a specific person raise complex questions about personal rights and work commitments. Legal experts emphasize that extracting behavioral logic—the patterns that define how a person works—blurs the boundary between legitimate work obligations and inalienable personal rights. As companies continue deploying digital clones of former employees, workers and legal scholars are calling for clearer protections around the ownership and use of individual expertise, personality, and identity data after employment ends.