Newshot Employee Fired for Using AI to Make PPTs, Court Orders Insurer to Pay Over 46,000 Yuan
A legal staff member at an insurance company used a third-party AI tool recommended by her supervisor to make PPTs and analyze data, and was dismissed by the company on the grounds of seriously violating information security rules. The court found that she had no intent to leak secrets, that the company had not explicitly prohibited AI use for work, and that it failed to prove any actual leak or damage. It ultimately ordered her reinstatement and compensation of over 46,000 yuan for lost wages and year-end bonus differences.
- The company could not produce evidence of a leak or damage
- Continue performing the contract and pay over 46,000 yuan
You hand internal company data to an AI platform recommended by your supervisor to make a PPT, and a month later you receive a notice terminating your labor contract. After hearing this labor dispute, the Beijing Chaoyang District Court held that the employee had no subjective intent to leak secrets, the company had not explicitly prohibited employees from using third-party AI, and it could not produce evidence of information leakage or actual damage. The ruling not only ordered compensation of over 46,000 yuan for lost wages and year-end bonus differences, but also required the insurance company to continue performing the labor contract.
The Company Recommended an AI Tool, and She Lost Her Job After Uploading a Spreadsheet
Ms. Wu used a tool recommended by her supervisor to process company spreadsheets, and a month later the insurance company terminated her contract. The court found the dismissal unlawful: a supervisor's recommendation does not mean she can hand over any file; and the company's inference of a leak based merely on the fact that "third-party AI retains data" is not enough to cost someone her job.
In July 2024, Ms. Wu worked as a legal staff member at an insurance company, handling compliance audits, risk verification, and related work. She uploaded a spreadsheet of company audit issue data that she had compiled and processed herself to a third-party AI platform to improve the efficiency of PPT creation and data analysis. This platform had been recommended multiple times by the company's department head and her direct supervisor.
A month later, the company issued Ms. Wu a "Notice of Termination of Labor Contract," citing her "serious violation of information security and related regulations," and simultaneously withheld that month's performance bonus and that year's annual bonus. Ms. Wu then applied for arbitration with the Beijing Chaoyang District Labor and Personnel Dispute Arbitration Committee. Neither party accepted the arbitration result, and the case proceeded to court.
During the trial, Ms. Wu said she uploaded the file to improve work efficiency, the tool had been recommended by her supervisors, and she had no intent to leak company secrets. The insurance company argued that third-party AI platforms collect and retain user data for model iteration, and that external uploading itself carries a risk of leaking secrets.
The court did not accept this presumption of risk. Although the insurance company had established confidentiality systems and information security management rules, and had signed confidentiality agreements with employees, it could not prove that the spreadsheet Ms. Wu uploaded constituted core confidential information as defined by law or by its rules, nor could it prove that the act caused information leakage, economic loss, or reputational damage. The ruling found that Ms. Wu had no subjective intent to leak company secrets, and that this was insufficient to establish that she seriously violated work discipline and should have her labor relationship terminated.
This dispute did not end with a single compensation payment. The first-instance ruling revoked the termination notice, required the insurance company to continue performing the labor contract, and ordered payment of lost wages and year-end bonus differences totaling more than 46,000 yuan. After the insurance company appealed, the second instance rejected the appeal and upheld the original ruling, which has now taken effect. The remaining question is: in the face of AI that has long since entered daily work, which boundary should companies actually write into their rules.
The Company Recommended the AI Tool, So Why Could the Employee Still Be Fired for Using It
What truly affects ordinary workplace participants in this dispute is whether a company can turn a vague confidentiality rule into a concrete boundary that employees know in advance and can check against afterward.
Ms. Wu had signed multiple confidentiality agreements and had received information security training. The company could require her to protect internal information; but the court found that its internal rules did not explicitly prohibit employees from using third-party AI tools (handing materials to external intelligent tools for processing), nor had it issued special rules or operating standards for AI work scenarios.
The company had also not conducted special training on the use of AI tools, nor told employees which files could be uploaded and which were strictly prohibited. The rules only said "information must be kept confidential," but did not explain what was permitted when using AI to process information. What employees faced was not an enforceable red line, but a set of prohibitions interpreted afterward by the company.
The platform Ms. Wu used was not something she found privately; it had been recommended multiple times by the company's department head and her direct supervisor. She put in the audit issue data spreadsheet she had compiled and processed herself, for the purpose of improving PPT creation and data analysis efficiency. A supervisor recommending a tool and a unit encouraging efficiency, on the one hand, and an employee handing internal materials to an unknown external platform on her own, on the other, are not the same in nature of risk.
Before every upload, employees face a concrete question: is the spreadsheet in hand public information, ordinary internal information, or protected information? If the employer writes only "do not leak" without specifying document classifications and tool boundaries, employees lack a basis for judgment even if they want to follow the rules. When companies rely on general rules to hold employees accountable after the fact, employees easily step over the line between improving efficiency and bearing risk.
The existence of management rules does not mean any risk can be directly escalated into termination of the labor contract. The aforementioned connection is not sufficient to establish a leak, and merely relying on the general rule that third-party AI platforms retain data to infer a risk of leakage is not enough to establish that an employee has reached the level of "seriously violating rules and regulations."
This case does not give a green light to all AI use at work. The insurance company can prohibit employees from uploading certain materials, can list which tools are permitted, and can set up approval, responsibility, and handling processes; but the rules need to be stated clearly first, and risk and responsibility also need evidentiary support. Managing AI cannot consist of just a vague line saying "pay attention to confidentiality"; how the mechanism is implemented and how employees should respond still need further discussion.
The remaining approximately 6 minutes are for members.
The free portion ends here. Behind the paywall there is also:
Paywalled highlight: The AI recommended by a supervisor ultimately became the reason the employee was fired?