Stay Neutral with AI-Drafted Workplace Allegations

As a workplace investigator, I recently received a multipage complaint that was tightly organized, meticulously framed, and peppered with legal terminology, presenting its conclusions with striking confidence. Yet when I sat down with the complainant to hear their account, a different picture emerged: The narrative was unfocused, key details were unclear, some allegations were walked back, and the underlying motivations shifted as the conversation unfolded. The contrast between the polished written submission and the disjointed interview raised an immediate question: What accounts for the gap between the complaint on paper and the story in person?

Experiences like this are no longer isolated. They highlight a growing tension in the investigative process, one that underscores, rather than diminishes, the core function of the investigator, because workplace investigators have never served as mere recorders of allegations. The role has always required translation: taking workplace narratives, often incomplete, emotional, or disorganized, and clarifying them into an identifiable scope. Investigators probe for detail, organize competing accounts, and assess credibility while maintaining neutrality toward the parties involved.

How-To Guide: How to Conduct a Workplace Investigation

The Rise of AI-Drafted Workplace Allegations

The rise of generative artificial intelligence (GenAI) has complicated this process in important ways. Increasingly, investigators receive complaints that are written with noteworthy legal sophistication. Submissions may include terminology such as “protected class,” “pretext,” or “retaliatory animus,” presented in structured narratives that closely resemble attorney-drafted pleadings. Yet when the complainant is interviewed, the individual may struggle to articulate the meaning of those terms, disclaim discriminatory motive, or materially retreat from assertions contained in the written complaint. Even more alarmingly, a complainant may deny that certain events occurred, raising concern that AI may have hallucinated a purported event. 

Determining if Someone Submitted an AI-Drafted Complaint

When an investigator suspects that a complaint was drafted with the assistance of GenAI, the investigator may address this suspicion directly, but their conversation should be handled with care. Questions about the drafting process should be framed in a neutral, matter-of-fact way, emphasizing that many individuals use tools or receive assistance when organizing their thoughts. For example, an investigator might ask whether the complainant used any writing aids, templates, or technology to help prepare the complaint, including AI tools. 

Related Article: Are Your Questions Undermining Your Investigations?

Maintaining a tone of curiosity rather than suspicion is essential. When framed respectfully and without judgment, such questions can encourage transparency and help ensure the interview focuses on the complainant’s personal experiences and recollections rather than the phrasing of the written complaint.

How Investigators Evaluate AI-Assisted Allegations

Ultimately, GenAI developments raise a central question: How should written allegations be evaluated when they appear to exceed the complainant’s own articulation of the underlying facts?

Investigators must walk a careful line. It is not the investigator’s role to advocate for any party or to help construct a theory that the complainant cannot independently express. At the same time, investigators cannot simply disregard written allegations when the complainant struggles to repeat them verbatim during an interview. The investigator’s obligation remains the same as it has always been — to understand what occurred in the workplace and evaluate the evidence objectively.

For that reason, investigators must remain disciplined in separating three related but distinct components of any complaint:

• The framing: the labels, headlines, or buzzwords used to describe the events.
• The allegation: what the complainant claims occurred.
• The evidence: the facts, observations, and documents surrounding that claim.

GenAI has the capacity to strengthen the first component without improving the latter two. That is to say, AI can produce polished framing without necessarily improving the factual narrative itself. As a result, the written complaint may emphasize labels and thematic language while the complainant’s own description of events remains more limited, nuanced, or differently expressed. The investigator’s task, therefore, is not to accept or reject the framing in a written complaint, but to determine whether the underlying factual narrative exists independently of that language.

Maintaining this distinction helps preserve investigative neutrality while ensuring that less articulate parties are not disadvantaged simply because they cannot express their experience in sophisticated terms. The investigator’s role remains active but disciplined: to clarify what the complainant experienced, to test the consistency of that account, and to evaluate the evidence without supplying the theory that makes the case.

Disciplined Focus

Ultimately, the emergence of generative AI in workplace complaints does not diminish the investigator’s role; it reinforces the importance of it. Written submissions may become more polished, more structured, and more sophisticated, but the investigator’s task remains grounded in the same principles that have always guided effective investigations: neutrality, careful questioning, and a disciplined focus on facts. By separating framing from allegations and allegations from evidence, investigators can ensure that the process remains fair to all parties while preserving the integrity of the investigative function. In an era where technology can shape how workplace concerns are expressed, the investigator’s role remains paramount. 

Rachel Reddick is an attorney investigator with Oppenheimer Investigations Group, headquartered in Oakland, Calif.