A labor dispute over AI-assisted work has ended with a ruling people will likely cite for years. According to a November 8 report by Workers' Daily, the Chaoyang District People's Court in Beijing recently finalized a case involving Wu, a legal officer at an insurance company. At her manager's recommendation, she used an AI tool to process internal documents — and was then fired for "seriously violating information security rules." The court saw it differently. It ordered the two sides to continue the employment contract and for the company to pay a year-end bonus difference, performance pay and wage losses, giving judges and employers a reference point for who bears responsibility when AI-assisted work goes wrong.
Encouraging employees to use AI, the report found, has become common practice. An employee at a Beijing bank described a dedicated AI working group, several mainstream large models pre-configured with agents for text processing and spreadsheet analysis, plus an internal "AI plus finance" bulletin. An internet company in Hubei goes further, giving programmers a monthly token allowance and reimbursing overages. Publishing professionals, for their part, routinely lean on AI for poster and layout design.
Case summary (per Workers' Daily)
Case: in-house counsel used AI on internal documents at a manager's recommendation, then was dismissed for "serious violation of information security rules"
Ruling: contract continues; company pays bonus difference, performance pay and wage losses
Reasoning: manager's recommendation, no explicit company prohibition, no compliance training, no actual leak — not "serious violation of internal rules"
Risk cases are piling up alongside the enthusiasm. In October, a reader spotted suspected AI prompt text left inside a printed novel. The same month, a popular-science book series was pulled from sale after AI-generated images turned out to include a two-headed shrimp. Zhou Yuan, an assistant professor at Beijing Institute of Technology's law school, warned that feeding unmasked client or financial data into third-party public models can leak trade secrets. Delivering unreviewed AI output, she added, can expose employers to breach or damages claims — and employees to dismissal.
So what should a careful employee do? He Ping, a law professor at Wuhan University of Technology, suggests checking corporate data-security rules first, then keeping written authorization such as emails or group messages when handling internal material. Zhou put the onus back on management, arguing that employers cannot push efficiency demands down while converting governance gaps into one-sided employee risk. Staff handbooks, in her view, should list permitted and prohibited AI scenarios, on the principle that "what the company has not made explicit in advance cannot be punished afterward."
The bigger ask is rulemaking. Both experts recommend dedicated judicial guidance or model-case references for AI-assisted office work. Until those arrive, the gap between companies that deploy AI and companies that also govern it will keep widening.