Can AI Investigate Workplace Bullying? What Companies Truly Need Is an "AI Investigation Assistant," Not an "AI Judge"

Generative AI is rapidly altering corporate management models, leading many HR managers to wonder: "Can workplace bullying and sexual harassment complaint cases also be handled by AI?"

My answer is: yes, and it will inevitably become an essential tool in HR management, including the investigation procedures for workplace bullying complaints. However, AI can only assist with investigations—it cannot replace them.

The idea that truly carries legal risk is not using AI tools, but treating AI as an investigation committee member or even using it to deliberate on whether a complaint is substantiated!

The correct answer to this issue lies not in whether to adopt AI, but in whether a governance model of "AI Assists, Humans Decide" is established.

For labor law compliance training, board member continuing education credits certification, external committee members for workplace bullying/sexual harassment, and litigation/mediation representation, please contact Yesin Law Firm (Phone: 02-25156822; Official Website: https://www.yesinlaw.com).

Keywords: AI, HR, workplace bullying

First: AI tools excel at administrative processes, significantly boosting investigation efficiency.

Keywords: Sexual harassment, complaint investigation, AI governance

Workplace bullying and sexual harassment cases often involve vast amounts of documentation, communication records, and interview notes. In the past, HR spent dozens of hours organizing these materials; today, AI can accomplish this rapidly.

From complaint intake, case classification, and risk grading to organizing LINE chat screenshots, emails, audio transcripts, and surveillance footage, AI can establish complete timelines, compile evidence, and issue reminders for statutory deadlines. It can even assist in drafting investigation report outlines and preliminary drafts—significantly lightening the data organization burden for staff so investigation committee members can focus their time on work that truly requires professional judgment.

Second: AI cannot replace legal judgment, nor can it replace evidence evaluation, fact-finding, and legal application in due process.

Keywords: Procedural justice, evidence, labor law compliance

The easiest mistake for enterprises to make is treating AI as an "electronic judge." Whether under the workplace bullying provisions of the Occupational Safety and Health Act or the investigation and handling procedures for workplace sexual harassment under the Gender Equality in Employment Act, the law ultimately requires enterprises to ascertain facts through lawful procedures rather than letting algorithms dictate outcomes.

AI can organize data, but it cannot judge which party's statement is more credible. It can analyze evidence, but it cannot determine whether workplace bullying or sexual harassment is substantiated, nor can it directly recommend reprimands, demotions, or disciplinary dismissals.

Conflict-of-interest recusal, hearing both sides, evidence admissibility, the principle of proportionality, and final legal evaluations must remain the responsibility of qualified dedicated handling personnel and investigation committee members.

Third: What enterprises truly need is an AI governance culture, rather than outsourcing all evaluation and judgment to AI.

Keywords: Personal data protection, corporate governance, legal risk, Occupational Safety and Health Act

While AI brings efficiency, it also introduces new legal risks—such as model hallucinations, algorithmic bias, unexplainable reasoning, and concerns regarding sensitive personal data or cross-border data transfers.

If a company inputs workplace bullying or sexual harassment case data directly into an uncontrolled AI platform, it may violate not only the Personal Data Protection Act but also confidentiality obligations surrounding complaint cases.

Therefore, when adopting AI, enterprises should establish a comprehensive AI governance framework—including data classification, access control, anonymization and de-identification, confidentiality measures, human review, and audit logs—so that AI becomes a tool to enhance governance quality rather than a new source of legal risk.

Attorney Chen Yexin’s Advice:

I believe that over the next five years, AI will completely transform how enterprises handle workplace bullying and sexual harassment complaint cases.

In fact, our firm currently utilizes AI tools—including NotebookLM—to assist in organizing evidence and party interview transcripts for complaint cases. However, we fully recognize that AI can never replace the ultimate judgments made by investigation committee members and dedicated handling personnel regarding evidence evaluation, fact-finding, and legal application.

A truly mature workplace bullying task force will not delegate the responsibility for final evaluation and judgment to AI tools. Instead, it will cultivate AI as a 24/7 investigation assistant that is well-versed in regulations, skilled at organizing evidence, and capable of highlighting procedural risks.

When AI handles data organization while humans handle judgment, and when technology boosts governance efficiency while legal frameworks safeguard fairness, enterprises can truly balance compliance, trust, and procedural justice.

Take a moment to review: when your company's HR department adopts AI tools today, are you building a governance mechanism that mitigates legal risks, or merely replacing humans with a faster "electronic investigator"—or even an "AI judge"—that cannot bear legal responsibility?