Newly leaked audio reportedly captures Mark Zuckerberg defending an internal program that tracked Meta employees’ Gmail, coding sessions, and internal tool usage to train AI models — on the same day 8,000 workers were being laid off. The recording was obtained and published by More Perfect Union, a labor-focused outlet.
What the Leaked Audio Shows
According to the report, the CEO is heard defending the data-gathering effort as necessary for AI competitiveness, citing the value of real-world developer and communications data for training. The program allegedly aggregated employee activity across Gmail, internal coding tools, and engineering workflows.
What makes the tape explosive isn’t just the surveillance — it’s the timing. Employees whose keystrokes, emails, and code reportedly helped train Meta’s AI reportedly found out about the round of 8,000 job cuts the same day.
The Data-Training Program
Advocates say the disclosures raise a simple question: did employees ever consent to their work product becoming training data?
- Email and messages scanned through Gmail tooling
- Coding sessions and internal tool traces captured
- Training uses beyond the stated internal-product purposes
Privacy and employment lawyers argue the program blurs the line between legitimate productivity analytics and the unauthorized repurposing of personal workplace data for AI.
The Same-Day Cutbacks
The layoffs — roughly 8,000 employees — were confirmed on the same day the audio surfaced, a coincidence that union organizers called a gut punch: the tools built with workers’ data, they charge, are part of the automation story behind shrinking headcounts.
Legal and Policy Fallout
Questions now center on:
- Whether employee consent was obtained for AI-training use
- Whether internal data boundaries were actually enforced
- Potential exposure under U.S. unfair-labor-practice and privacy statutes, and under EU GDPR-style obligations in Meta’s international operations
Labor advocates point to this case as a reason workers need stronger voice in how AI is deployed — and whose data pays for it.
What This Means
For Meta, the leak lands at the heart of two of tech’s most sensitive fault lines: the ethics of AI training data and the human cost of AI-driven labor cuts. Even if no law was broken, the story — training models on the transcribed work lives of employees who will be laid off — could follow Meta through labor disputes, shareholder pressure, and regulatory reviews for a long time.