AI Regulation Tracker / Guidance in drafting
Taiwan's Ministry of Labor drafts a workplace AI guideline on hiring, appraisal, and human review
Taiwan's labor regulator is preparing guidance that would tell employers to notify job applicants when AI screens them, keep a human making the final call, and be able to explain the result. It is still in drafting, so nothing binds yet.
Taiwan's Ministry of Labor is drafting guidance on how companies bring AI into the workplace, and early signals point to three recurring duties: tell applicants when AI is screening them, keep a human making the final decision, and be able to explain how an AI reached its result. None of this is in force. The instrument is still being written, and even its title is not confirmed.
What is on the record is limited but specific. Deputy Minister Li Chien-hung discussed the intended scope on March 23, 2026, in reporting carried by Taiwan's Central News Agency and United Daily News. The document has been referred to as a "Guideline for Enterprises Introducing and Using AI" (事業單位導入及使用人工智慧指引), though that title is unverified and could change before publication. Readers should treat everything below as a preview of what may come, not a rule to comply with today.
What the guideline is expected to require
As reported, employers that use AI to recruit would need to notify applicants in advance that AI is part of the process, and a human would make the final hiring decision rather than the system. For current staff, AI-driven performance appraisals would have to be disclosed to the worker, reviewed by a human, and explainable, meaning the employer can articulate why the AI produced a given assessment. The reported aim is to guard against algorithmic discrimination in employment and to protect worker-data privacy while keeping a person accountable for the outcome.
The Ministry has framed the scope broadly. According to the same reporting, the guideline is meant to touch employment discrimination, privacy, labor relations, occupational safety and health, and vocational training. That breadth suggests the Ministry sees AI reaching beyond hiring into how work is supervised, measured, and trained, though the operational detail for each area is not yet public. It is worth being precise about what "explainable" appears to mean here: not that the employer publishes the model, but that a person can give a worker a plain-language account of why an appraisal came out as it did. On the reported framing, the human is not a rubber stamp on an AI result but the party who owns the decision and can defend it.
Why the Ministry is moving now
The reported driver is adoption. Roughly one-third of Taiwan firms are said to be considering AI in recruiting, per the secondary reporting, which puts screening and ranking tools in front of a large share of the labor market before any dedicated rule exists. A notice-and-human-decision model is the Ministry's apparent answer: let employers use the tools, but require that a person can be pointed to as the decision-maker and that a rejected applicant is not quietly filtered out by a system no one will explain. Guidance also tends to arrive before legislation in Taiwan's approach to AI, setting expectations that sector rules or self-regulation later harden, so a published version would signal direction to employers well ahead of any binding duty.
What this does not do
This is not a statute, and on the current record it would not be one. A published guideline of this kind would be administrative guidance, not a binding law with its own penalties, so the immediate consequence of the reported document is direction rather than fines. It does not yet impose an audit requirement, a filing, or a certification. And because the text is unpublished, any specific obligation, threshold, or effective date described here should be read as reported intent that could shift. Employers should not restructure hiring around a rule that does not yet exist; the sound move is to prepare, not to comply with a draft.
The cross-border read for US firms
For a US employer or an in-house team watching from abroad, Taiwan's approach rhymes with rules already live elsewhere. New York City's Local Law 144 requires bias audits and candidate notice for automated employment decision tools. The EU AI Act treats AI used in recruitment and worker management as high-risk, with transparency and human-oversight duties. Illinois' Artificial Intelligence Video Interview Act requires notice and consent before AI analyzes a video interview. Taiwan's reported notice-plus-human-decision-plus-explainability model is the emerging APAC entry in that same pattern. A US firm hiring in Taiwan would eventually fold this into the same global playbook it already runs for New York, Illinois, and the EU. For now, the practical value is early warning: the direction of travel is consistent across jurisdictions, even where the Taiwan text is not final.
Frequently Asked Questions
What is Taiwan's Ministry of Labor changing about workplace AI?
It is drafting a guideline, reported as a "Guideline for Enterprises Introducing and Using AI," that would ask employers to notify applicants when AI screens them, keep a final human decision in hiring, and disclose, human-review, and explain AI performance appraisals. It is not published and not in force.
Who would be affected if it is issued?
Employers in Taiwan that use AI to screen or rank applicants or to appraise existing staff, along with their HR, compliance, and employment-law teams. Reporting suggests about one-third of Taiwan firms are already considering AI in recruiting.
Is this a binding law with penalties?
No. On the current record it is administrative guidance in drafting, not a statute. A basic caveat: the exact title, scope, and publication date are unverified, and the document could change before release.
What should employers do now?
Map where AI already influences hiring and appraisal decisions, and monitor the Ministry of Labor's site for the published text. Building an applicant-notice step, a documented human decision, and an explainability record now makes later compliance a lighter lift.
Sponsored Training
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.