A newly filed lawsuit in the Superior Court of New Jersey, Gloucester County, offers a cautionary tale for every employer using AI-powered meeting tools. In her Complaint, the plaintiff alleges that the employer’s AI note-taker took notes of her termination meeting and then emailed her the transcript of what her former supervisors said after she left the meeting.
The plaintiff’s claims have not been adjudicated, and the allegations remain unproven, but the scenario they describe serves as an important reminder for HR professionals, in-house counsel, and managers that AI tools can capture and distribute information in ways their users never intended.
The plaintiff’s claims
According to the Complaint, the employee was hired in November 2024 and terminated less than three months later during a video conference on February 3, 2025. Three company representatives joined the video conference and informed her that it was not “working out,” ending the call abruptly.
Before the call began, an administrator had activated an AI meeting assistant that records and transcribes meetings. The employee alleges that after she disconnected, the AI tool continued recording. During the post-termination discussion, a manager allegedly described the ideal replacement candidate as “hopefully a relatively strapping young man.”
The AI meeting assistant then reportedly emailed the employee a link to the full transcript, including the remarks made after she left. The employee has since filed suit under the New Jersey Law Against Discrimination (NJLAD), alleging that her employer terminated her employment based on her gender.
Why this matters for employers
Setting aside the merits of the underlying discrimination claim, the Complaint identifies three avoidable vulnerabilities that any employer could face:
- An AI-generated transcript distributed to the wrong audience. The tool automatically shared a transcript with a meeting participant who was also the subject of the termination discussion. Employers often have no control over AI distribution lists once a tool is activated.
- A vague termination explanation. “Not working out” is an invitation for a plaintiff’s attorney to supply an alternative narrative. Without documented, specific reasons, the employer has little to counter a claim that the real motivation was discriminatory.
- A replacement discussion using demographic language. Describing the ideal successor by sex and physical build—rather than by qualifications—can transform routine staffing conversations into powerful evidence of discriminatory intent.
Beyond these immediate risks, AI meeting tools raise broader concerns about attorney-client privilege and document retention. If counsel participates in a call being recorded by a third-party AI service, the resulting transcript may not be privileged—and could become discoverable in litigation. Organizations may also wish to evaluate whether existing record-retention policies adequately address AI-generated transcripts and summaries.
Practical takeaways
- Avoid using AI note-takers in termination meetings or calls involving counsel. Automated recording adds minimal value to these sensitive conversations while creating a detailed, distributable record. Consider documenting termination decisions through standard HR paperwork instead.
- Adopt an AI meeting-tool policy. Consider a policy that identifies who may activate recording and transcription tools, specifies which types of meetings are off-limits, addresses notice and consent requirements, and controls how transcripts are stored, retained, and distributed.
- Provide clear, specific, and documented termination reasons. It is advisable for decision-makers to articulate a consistent explanation supported by contemporaneous records. Vague rationales leave gaps that opposing counsel will fill.
- Keep replacement discussions focused on job qualifications. Comments about a preferred candidate’s sex, age, appearance, or physical characteristics should be avoided in hiring conversations. Practitioners should train managers to discuss skills, experience, and role requirements—nothing more.