Job Media and Scouting Services: The Boundary with Employment Placement and Personal Information
Hello, I'm Noriaki Asato, Representative Attorney at LegalAgent.
In consultations about job media, scouting services and recruitment AI, whether the company's own service falls under employment placement is often at issue. Even if the operator sees itself as "just posting job information" or "just providing a place where companies and candidates exchange messages directly," depending on the features built into the service and how it is run day to day, it may be found to require a permit for an employment placement business.
In its Distinction Between Provision of Recruitment Information and Employment Placement (Japanese), the Ministry of Health, Labour and Welfare sets out the criterion that acts of mediating the establishment of an employment relationship between job offerors and job seekers constitute employment placement. From the stage of planning and designing a service, it is important to look carefully at how far the system and staff intervene in matching.
The Distinction Between Employment Placement and Provision of Recruitment Information
Employment placement refers to the act of accepting applications for job offers and job seeking and mediating the establishment of an employment relationship between job offerors and job seekers. In contrast, provision of recruitment information is limited to the act of providing job information or job seeker information. If a job media service posts job information and job seekers apply based on their own judgment, the service tends to be easier to classify as provision of recruitment information.
However, if the service operator screens candidates and recommends them to specific companies, modifies the content of communications between the parties, or becomes involved in negotiating terms rather than merely coordinating interview schedules, mediation is likely to be recognized and the service is more likely to be regarded as employment placement. Even with an AI-based recommendation feature, if candidates and job offers are in substance screened and matched according to the operator's criteria, the legal characterization must be checked carefully.
Under the Ministry of Health, Labour and Welfare's criteria, whether a series of processes is performed manually or automatically by a program is not the deciding factor. Note that even automated processing by AI can constitute employment placement. In practice, the features of the service as a whole are checked from the following perspectives.
- Check whether the service accepts applications for job offers from hiring companies and applications for job seeking from candidates
- Distinguish whether the companies and candidates decide the scope of information presented, or the operator screens it independently
- Verify whether the service merely reorders the display, or adds new evaluations such as the likelihood of being hired
- Understand whether companies and candidates contact each other directly, or the operator is involved in the content or channel of communications
- Check that advertising, sales materials, terms of use and actual operations do not promise candidate introductions or successful hiring
For example, if job seekers can view all job offers and the service merely reorders the list according to their past viewing history, this is said not necessarily to constitute modification of information. On the other hand, a design that assigns each candidate a proprietary score such as expected annual salary or likelihood of receiving an offer, and displays only a limited set of job offers based on the operator's judgment, tends to strengthen the character of mediation. It is important not to judge from the specifications of a particular screen alone, but to check the entire transaction flow from user registration through to billing.
Designing Scouting Features and AI Recommendations
In scouting services, a common design is for companies to search for candidates in a database themselves and send messages directly. In determining the boundary with employment placement for this design, you need to establish whether the operator merely provides a search environment and messaging feature, or is involved in selecting candidates and making individual recommendations. Check how candidate information is collected and how consent to provision to third parties is obtained, whether companies narrow down candidates themselves, and whether the operator determines the order of recommendations using its own scoring.
In addition, a success-fee pricing structure under which fees arise when a hire is made, or scheduling and negotiation of terms by the operator's staff, tends to make the boundary with employment placement a point of debate. However, a success-fee pricing mechanism alone does not automatically mean that the service is employment placement; the degree of actual screening of candidates and intervention in communications is assessed comprehensively. Look specifically at whether the operator merely relays schedule communications mechanically or intervenes in aligning terms or persuading the parties.
When introducing AI recommendations, the focus is not on the complexity of the algorithm but on who specifies the recommendation criteria and what information is excluded. Also distinguish whether the service merely changes the display order or makes other information unviewable. The legal assessment differs between a case where the hiring company specifically designates criteria such as qualifications held and work location and information is delivered mechanically in line with that designation, and a case where the operator, using its own criteria, limits the candidates shown to hiring companies or the job offers shown to job seekers.
Even where the screen displays "recommended," scores such as the likelihood of applying, the probability of receiving an offer or salary predictions may be used behind the scenes. Without understanding the evaluation criteria, training data and conditions for excluding items from display, you cannot address not only whether the service constitutes employment placement but also trouble arising from unfair bias or erroneous recommendations. It is advisable to state clearly in the specifications the input information, screening conditions and the extent to which manual intervention is possible, and to put in place an operation that rechecks, whenever the AI's decision logic is updated, that the premises of the legal analysis still hold.
Protecting Job Seeker Information and Notification of Specified Recruitment Information Provision
Job media and scouting services may handle names, work histories, desired salaries and screening results, and even sensitive information such as health conditions and reasons for leaving a previous job. With respect to the Act on the Protection of Personal Information, check the specification of the purpose of use, procedures for provision to third parties, the setting of retention periods and security control measures. It is important to design screens so that users can clearly understand "which information will be disclosed to which companies, and when."
Also note that a business that provides recruitment information by collecting the personal information of persons seeking to become workers must file a notification with the Labor Bureau as a specified recruitment information provision business, even when it does not constitute employment placement. The notification obligation and the various operational guidelines under the Employment Security Act apply separately even to services that do not constitute employment placement.
The fact that a candidate has registered a profile on the service cannot be regarded as consent to unlimited provision of information to every company. The behavior of the actual screens and the statements in the privacy policy must be consistent on points such as whether candidates can choose the scope of disclosure of their information themselves, whether there is a mechanism to obtain consent before disclosing the person's name and other details to a hiring company, and how search results and recommendation histories are managed after the user withdraws.
Where recruitment AI infers a person's intention to change jobs or their characteristics from what is written in a résumé and generates new evaluation data, the purpose of use and accuracy of the inferred information also come under consideration. Keeping logs in preparation for future verification or investigation of recommendation reasons is useful, but retaining all data permanently after withdrawal and, conversely, deleting everything immediately across the board each carry legal and practical risks. It is advisable to organize the necessary retention periods and access permissions according to the purpose of use and maintain unbiased operations.
Accurate Display of Job Information and Operational Management
In operating a job media service, efforts to keep the posted job information accurate are also essential. Check whether wages include fixed overtime pay, and what the working hours and type of employment are. If the displayed terms, including work location and whether remote work is possible, differ from reality, disputes with job seekers may arise. Providers of recruitment information are legally required to take measures to display job information accurately and keep its content accurate and up to date.
In practice, simply having hiring companies represent and warrant in the terms of use that they will "enter accurate information" may not be regarded as fulfilling the operator's responsibility. Decide in advance on a contact point for receiving reports from job seekers, procedures for fact-checking, criteria for suspending postings, and criteria for promptly deleting job offers after recruitment closes and duplicate postings.
Particularly where the service has features in which AI summarizes job postings or automatically formats job titles and terms, watch for the risk of wording that the original hiring company did not specify being introduced. To prevent displays that make the base salary look high when fixed overtime pay is included, or modifications that mislead readers into thinking a position is employment when it is in fact a service agreement, it is effective to compare content before and after generation and to display key items in fixed form.
Even if the terms of use state that "the Company does not engage in employment placement," the legal assessment will not change if the reality is different. If sales materials say "we carefully select and introduce the best talent for your company," customer success staff individually match candidates, or the contract takes a form in which only success fees are received, check whether the external explanation is consistent with reality.
In legal review, it is effective to examine not only the text of the terms of use but also the actual product screens, sales materials and operational manuals, laid out in line with the flow of use. Particularly where recruitment AI is built in, adjustments to the recommendation logic or revisions to the pricing structure can change the degree of involvement in screening information and in communications. It is practical to adopt a mechanism for checking at every specification change, rather than stopping at the judgment made at the initial release.
Services connected to this topic