Behind BOSS Zhipin's Monthly Active Users Exceeding 70 Million: What Legal Risks Does the Full AI Recruitment Process Involve? Lawyer Interprets Key Compliance Points
Lead: When AI becomes the recruitment 'examiner,' how can the law keep up?
BOSS Zhipin's Q2 2025 financial report shows that its monthly active users (MAU) surpassed the 70 million mark for the first time, while AI services have expanded from resume screening to the entire recruitment process, including interviews and salary negotiations. Behind these figures lies the reality that AI technology is profoundly reshaping China's human resources market. However, as algorithms decide who gets an interview, AI interviewers evaluate candidate performance, and systems automatically generate hiring recommendations, a series of legal issues have also surfaced: Does algorithmic recommendation constitute employment discrimination? Is AI processing of massive volumes of resumes compliant? How can job seekers seek legal redress when faced with unfair treatment?
As a lawyer deeply engaged in the field of labor law, I believe that while technology races ahead, legal boundaries must remain clear. This article will draw on the Personal Information Protection Law of the People's Republic of China, the Employment Promotion Law of the People's Republic of China, the Provisions on the Administration of Algorithmic Recommendations in Internet Information Services, and other laws and regulations, and integrate judicial practice, to break down for job seekers and employers the legal risks and compliance pathways throughout the entire AI recruitment process.
1. Algorithm Screening of Resumes: The "Hidden Trap" of Employment Discrimination
The most basic application of AI in recruitment is resume screening. The system automatically filters candidates by analyzing dimensions such as keywords, education, and work experience. However, the problem lies in the fact that if the algorithmic model is trained on historical hiring data, it is highly likely to use sensitive characteristics such as gender, age, and region as implicit screening criteria. For example, an AI system at a technology company once automatically downgraded scores on women's resumes because of the high proportion of male engineers in historical data — this directly contravenes the prohibitive provision of Article 3 of the Employment Promotion Law: workers shall not be subject to discrimination in employment on account of ethnicity, race, gender, or religious belief.
What is more challenging is that algorithmic decisions are often opaque. After a job applicant is rejected, it is difficult to know the specific reasons. Under Article 12 of the Provisions on the Administration of Algorithmic Recommendations in Internet Information Services, algorithm recommendation service providers shall provide users with options that do not target their personal characteristics, or provide users with convenient options to disable algorithm recommendation services. However, this provision mainly targets information content services, and its applicability to recruitment scenarios remains contentious.
In practice, I advise job seekers to retain evidence such as application records and system feedback screenshots. If they suspect algorithmic discrimination, they may file complaints with labor inspection authorities or bring litigation on equal employment rights. In 2022, a court in Beijing ruled in an employment discrimination case triggered by AI screening, finding that the company constituted infringement. For enterprises, they should regularly audit algorithmic models to ensure they do not contain discriminatory characteristics, and clearly disclose screening criteria in recruitment rules to avoid "black box" operations.
II. AI Interview Stage: Compliance Red Lines for Personal Information Collection
AI interviews have become standard practice in campus recruitment at major tech companies. The system conducts comprehensive scoring by analyzing candidates' micro-expressions, voice intonation, and interview responses. In this process, companies may collect biometric information far beyond what is necessary for the position. According to Article 28 of the Personal Information Protection Law, biometric information constitutes sensitive personal information and may only be processed when there is a specific purpose and strict necessity, and when stringent protection measures are implemented.
In reality, many companies neither clearly inform job applicants of the scoring logic used in AI interviews nor obtain separate consent. This violates Article 29 of the Personal Information Protection Law: the processing of sensitive personal information shall be subject to the individual's separate consent. In addition, if an AI interview system stores data on overseas servers, it may also trigger the cross-border provision rules under Article 38 of the same law.
For job seekers, if they discover that an enterprise has not disclosed its use of AI interviews, they may exercise their rights to access and copy under Article 50 of the Personal Information Protection Law against the personal information processor, and require the enterprise to explain the processing rules. If a job opportunity is missed due to an AI misjudgment, it is currently difficult to directly claim compensation in judicial practice; however, in conjunction with Article 1037 of the Civil Code, one may request the correction of erroneous information.
Enterprises should establish a full-process compliance mechanism: obtain separate consent before interviews, disclose the algorithm logic during interviews, and delete raw data within a set period after interviews. In 2024, a foreign enterprise in Shanghai was fined 500,000 yuan for illegally collecting emotional data through its AI interview system—a case worth heeding.
III. AI Salary Negotiation and Hiring Decisions: Algorithms Cannot Replace "Human Judgment"
The latest trend is AI's involvement in salary negotiations and final hiring decisions. The system automatically generates offer proposals based on market data, candidates' salary expectations, and the company's internal compensation structure. This may appear efficient, yet it could trigger a chain of issues under labor law.
First, if AI-generated salary proposals fall below legal standards, such as below the local minimum wage, the enterprise still bears legal liability. Article 48 of the Labor Law clearly stipulates that the state implements a minimum wage guarantee system. Algorithmic recommendations cannot serve as an excuse for violating mandatory provisions. Second, AI decision-making may overlook special circumstances of candidates, such as reasonable accommodation needs for persons with disabilities, thereby violating the equal employment requirements of Article 38 of the Law on the Protection of Persons with Disabilities.
The deeper issue is whether an AI decision to reject a candidate constitutes an alienation of the principle of "equality, voluntariness, and mutual consultation" as stipulated in Article 3 of the Labor Contract Law. This principle emphasizes that the conclusion of a labor contract should be based on the genuine expression of intent from both parties. If one party (the enterprise) relies entirely on algorithms, it may bypass the consultation process. In my view, enterprises should retain a manual review step to ensure that the final decision is made by qualified HR personnel.
The EU's Artificial Intelligence Act of 2023 has explicitly prohibited the use of social scoring AI in recruitment. Although China has not yet issued dedicated regulations, documents such as the Measures for the Labeling of AI-Generated Synthetic Content are being refined. Enterprises should monitor regulatory developments and position AI as an assistive tool rather than a decision-making entity.
IV. Job Seeker Rights Protection Guide: How to Legally Respond to AI Unfairness
In the face of unfairness in AI recruitment, job seekers are not helpless. First, preserve evidence, including application records, screen recordings of AI interviews, and system rejection notices. Second, file complaints with the platform. Platforms such as BOSS Zhipin, pursuant to Article 38 of the E-Commerce Law of the People's Republic of China, bear management responsibility for the illegal acts of operators within the platform, and may be required to remove discriminatory job postings or suspend services for the relevant enterprises. Third, file administrative complaints. Report the employer to the labor inspection team for violating the Employment Promotion Law, or report algorithmic violations to the cyberspace administration authorities. Fourth, initiate judicial proceedings. A privacy lawsuit may be filed based on Article 1032 of the Civil Code, or an equal employment rights dispute may be filed pursuant to Article 62 of the Employment Promotion Law.
In 2024, a job seeker in Guangzhou sued a company for infringing on their right to personal information because the AI interview system did not disclose the purpose of recording, and was ultimately awarded emotional distress compensation. This proves that legal remedies are effective. For lawyers, the key to such cases lies in proving causation—that is, the direct link between algorithmic decision-making and the resulting harm.
V. Enterprise Compliance Recommendations: Empowering Recruitment with AI Within the Framework of the Rule of Law
For companies currently using or planning to introduce AI-based recruitment, my recommendations are: First, conduct an algorithmic impact assessment. Refer to Article 55 of the Personal Information Protection Law to assess in advance the risks of processing sensitive personal information. Second, establish transparent rules. Clearly disclose to job seekers the scope of AI use, data purposes, and appeal channels. Third, establish a human intervention mechanism. Ensure that algorithmic recommendations do not directly determine hiring outcomes, and retain records of manual approval. Fourth, regularly train the HR team. Enable practitioners to understand the limitations of AI and legal red lines.
Guangdong Zhiming Law Firm's Labor Law Team has long focused on the intersection of technology and human resources. We have assisted numerous enterprises in completing compliance renovations of AI recruitment systems and provided rights-protection support for workers facing algorithmic discrimination. Technology knows no boundaries, but the law has its bottom line. In an era where AI is reshaping the recruitment ecosystem, only by upholding compliance as the baseline can we achieve a win-win outcome of both efficiency and fairness.
If you or your company are facing legal uncertainties related to AI recruitment, please feel free to contact us at any time for professional legal advice.
