Overview
- Meta paused its Model Capability Initiative after late-June reporting and an internal high-priority security incident prompted the company to investigate the program’s data practices.
- The program, launched in April, logged fine-grained workplace interactions including mouse movements, clicks, keystrokes, full prompts, transcriptions and some private conversations, and internal documents showed that this material was broadly accessible to staff.
- A Meta spokesperson said the company has found no evidence so far that any data was improperly accessed and that the initiative was designed with privacy safeguards while the investigation continues.
- At the same time, Meta is moving ahead with leadership changes at WhatsApp, with Kunal Shah taking charge, and third-party firms such as DoubleVerify are expanding services to Meta’s ad platforms, placing product and data-access issues under greater commercial and regulatory scrutiny.
- If the investigation finds access-control failures, the likely outcomes include tighter internal controls, potential regulatory action in the EU over messaging data access, and increased limits on how Meta uses employee and message data for AI development.