An employee submits a voicemail as proof her manager threatened her job over a weekend shift. It sounds exactly like him. He swears he never left it. Forensic voice analysis can't settle it either way — and both sides have already lawyered up. Fabienne Lopez joins HR Voices to walk through what happens when HR has no playbook for evidence that might not even be real, and has to build the investigation in real time.

What you'll hear in this episode:

Why jumping straight to "did the manager do it or not" is the riskiest move you can make in a case like this
How to bring IT in as a real investigative partner — not just for AI detection, but for carrier records, IP data, and file history
A framework for treating AI-generated evidence as just one more data point, not a special category that changes your process
Why the relationship between the two parties usually matters more than who's "right" — and what it means when someone skips a direct conversation and goes straight to a complaint
The ethics of asking to see someone's phone as evidence, and where that request gets complicated fast
How to keep a witness list contained without it turning into a witch hunt
Why "we'll never know for sure what happened" is sometimes the actual outcome — and what HR does next when the evidence never resolves cleanly

About Fabienne Lopez:

Fabienne is Chief People Officer at Xponential Fitness, where she leads people strategy across the company's portfolio of fitness brands. With a background spanning HR leadership roles across multiple industries, she specializes in employee relations, investigations, and building HR functions that operate as true business partners rather than compliance gatekeepers.

The Deepfake Voicemail: When AI Makes Evidence a Liar

HR Voices

The Deepfake Voicemail: When AI Makes Evidence a Liar