Meta Halts Employee Data Tracking After Sensitive Info Reportedly Exposed

Meta has abruptly halted a controversial internal program that tracked employee communications and data to train its AI models, following reports that sensitive personal information — including health and financial data — was exposed. The decision marks a rare retreat for the tech giant, which had been quietly using the program to feed its growing artificial intelligence ambitions.

The program, which Meta declined to name publicly, involved collecting and analyzing messages, meeting transcripts, and other internal data from employees across its global workforce. Insiders say the goal was to improve Meta's AI models, particularly for natural language processing and workplace automation tools. However, a report from a whistleblower inside the company revealed that the system inadvertently captured highly sensitive details — such as medical records, banking information, and performance reviews — and stored them without adequate safeguards.

Meta's CEO Mark Zuckerberg has long emphasized AI as a core priority, pouring billions into infrastructure and talent. The company uses vast datasets from its platforms — Facebook, Instagram, and WhatsApp — to train models like Llama. But employee data offered a new, supposedly controlled source for refining internal tools. Now, that approach has backfired, reigniting debates about workplace surveillance and the ethics of using worker data for AI.

The halt came after a group of employees formally complained to Meta's ethics committee, and CNET reported the exposure last week. According to the source, the program was active for at least six months before the shutdown. A Meta spokesperson told reporters that the company "takes data privacy seriously" and is reviewing its practices, but did not confirm whether disciplinary actions would follow. The controversy has drawn attention from the U.S. Equal Employment Opportunity Commission and at least one state attorney general, who are considering inquiries.

Privacy experts warn that this incident reflects a broader trend in Silicon Valley: companies will exploit any available data — including internal employee data — to train AI without explicit consent. "This is a canary in the coal mine," said Dr. Lisa Wang, a researcher at the AI Ethics Lab. "If Meta can't protect its own employees, what hope do regular users have?" The scandal also threatens to undermine trust in Meta's AI initiatives, especially as it races against OpenAI and Google to lead in generative AI.

Moving forward, Meta faces a dual challenge: restoring internal confidence and avoiding regulatory penalties. The company has promised a full security audit and will consult with employee representatives before resuming any similar program. Several shareholder groups have called for an independent third-party review. The outcome of these investigations could set new precedents for how tech companies handle private data — both from users and from the very people building their products.