What to Know
- AI-driven workplace complaints demand deeper fact-checking beyond polished legal language.
- AI can improve misconduct reporting while increasing risks of inaccurate legal claims.
- Internal investigations must prioritize facts, credibility and AI governance safeguards.
In the age of artificial intelligence, investigators handling workplace complaints, ethics hotline reports and other reports of misconduct are increasingly seeing a shift: More and more of the complaints crossing their desks appear to have been drafted, at least in part, with the assistance of AI.
A practice that was uncommon only recently is quickly becoming routine. On the surface, these submissions can look as though they were prepared by an attorney, with numbered allegations, references to exhibits and evidence, and specific demands for relief. The language is often polished, formal and confident in tone.
A complaint that once might have been a few plainly worded paragraphs can now run dozens of pages, cite multiple legal theories, and assert that particular conduct violates specific company policies or laws. AI has effectively given many complainants access to something that can sound like a lawyer — but not necessarily a good one.
AI-generated legal analysis can be wrong or misplaced, and beneath the polished complaint, the same questions of fact and credibility remain. The challenge for companies, lawyers and human resources professionals who handle these complaints is how to adapt to this new reality.
Hallmarks of an AI-Assisted Complaint
If you ask someone who regularly conducts internal investigations whether they can tell if a submission was drafted with the assistance of AI, most will probably tell you they know it when they see it.
An AI-assisted complaint is often well written, longer than necessary, highly structured and more formal than a typical complaint or whistleblower report. It may feature numerous headings, legal terminology, distinctive formatting and broad conclusions that are not supported by specific facts.
AI’s influence on investigations extends beyond the initial complaint. The complainant and witnesses may use AI to prepare for interviews, document their thoughts and recollection of key events, or prepare written statements. They also may use AI notetaking tools to transcribe their interview, potentially without the interviewer’s knowledge.
Understanding the Risks
The central risk that AI poses to internal investigations is not that AI-assisted complaints are dishonest, or even that there is anything inherently wrong with someone using AI to prepare a complaint or report of misconduct. Companies should encourage reporting — in whatever form their employees are most comfortable providing it.
Indeed, for employees, AI can be quite helpful to clarify prose, organize ideas and present useful background information. It can also help employees communicate across language barriers and make it easier to describe difficult experiences. It can be a powerful research tool, as well.
All of this can make reporting legitimate concerns more accessible, which companies should welcome.
There are advantages to employees having access to a generative AI tool when they might not have access to an attorney for a variety of reasons, including cost. But AI is not a substitute for legal advice and judgment. AI tools are built to produce fluent, persuasive writing, not to verify facts or evaluate the merits of a legal claim, and as a result, they may give a complainant unwarranted confidence that a particular claim is legally sound.
The risk for companies and organizations is simply that investigators are not trained and prepared to handle this new type of complaint.
Beyond identifying a complaint that was drafted with AI, the first concern for an investigator evaluating such a complaint is that polished writing and confident conclusions can make a complaint appear stronger on its face than it really is.
AI tools may fill gaps, smooth over ambiguity or present uncertain information as settled without clearly signaling that they have done so. An AI-assisted complaint may also draw inferences that the facts do not support or conclude that a policy or law was violated based on a misunderstanding or disregard of the applicable standard.
In addition, although an AI-assisted complaint does not inherently lack credibility, something can be lost when an AI tool substantially shapes the account of the individual who witnessed or experienced the reported conduct. A complaint written in the individual’s own words may provide useful context about how the person understands the events, which issues they consider most significant, and whether they are reporting firsthand knowledge, an inference or information obtained from others.
Some of that context may be obscured when AI substantially shapes the narrative.
How Companies and Investigators Can Adapt
Given AI’s influence, the fundamentals of a thorough investigation — documents and interviews — become even more important.
Investigators should take extra precautions not to be swayed by a polished and forceful complaint. With AI, confidence does not necessarily equate to the strength of the underlying support. Investigators should be more vigilant about collecting and analyzing documents and other information to verify the factual assertions in a complaint or report.
Interviews are even more critical for testing recollections and determining whether the account in an AI-assisted complaint is consistent with the complainant’s own memory. Investigators should avoid giving undue weight to the written submission, or forming a bias for or against the allegations, when the language may not be the complainant’s own.
They should distinguish carefully between what the individual personally observed, what the individual inferred and what AI may have introduced or suggested. Detailed, fact-specific questions are essential to making those distinctions and assessing credibility.
Investigators should also be mindful of the source of materials a complainant or witness provides in the course of an investigation. Contemporaneous notes can be a useful data point, but investigators should confirm when and how the notes were created. Did the individual draft them in real time or immediately after a key event, or use AI later to compose or reshape the narrative?
The use of AI does not necessarily render notes unreliable, but investigators should understand when the notes were created, what source material was provided to the tool and how extensively the tool shaped the final account.
With the advent of AI notetaking tools, investigators should also consider the confidentiality and attorney-client privilege concerns that may arise when an interview is transcribed by either the witness or the interviewer.
Investigators can also use AI to their advantage. Subject to applicable confidentiality, privilege, security and other organizational requirements, AI may help investigators work more efficiently, reduce administrative and repetitive tasks, and focus their time on judgment-intensive work.
For example, AI may assist with preparing interview outlines, reviewing documents, drafting summaries and building chronologies. But it cannot replace an experienced investigator’s judgment in assessing credibility, weighing evidence, or applying the relevant legal or policy standard.
Most importantly, given the quickly changing landscape, companies, lawyers and human resources professionals tasked with conducting internal investigations should open a dialogue with their colleagues about how AI is affecting their practice and how they can support one another and adapt.
Companies should establish clear protocols addressing approved tools, permitted and prohibited inputs, human review of AI output, confidentiality, privilege, and data security, as well as procedures for AI notetaking or transcription.
Training should help investigators identify AI-assisted complaints, adjust their investigative techniques and reinforce best practices that apply regardless of how much AI is involved.
Conclusion
AI is changing how workplace complaints and reports of misconduct are prepared and presented, but it has not changed the investigator’s core task. Whether a complaint is drafted by an employee, an attorney or with the assistance of AI, investigators should look beyond the form of the submission and focus on the underlying facts.
That means evaluating allegations through documents and interviews, assessing credibility, and applying sound judgment to the facts and information gathered.
As AI tools continue to evolve, companies should ensure that investigators understand both their risks and their potential uses. The goal is not to resist the technology, but to adapt investigative practices so that the process remains fair, thorough and grounded in fact.