Meta has suspended its internal Model Capability Initiative, a program that tracked employee computer activity to train artificial intelligence models, following significant staff backlash and data security concerns.

Key Points

• The Model Capability Initiative tracked employee keystrokes and mouse clicks to gather data for training Meta’s artificial intelligence models.
• Meta paused the program on Monday after discovering that some collected data was potentially accessible to unauthorized internal staff.
• Nearly 2,000 employees signed a petition opposing the surveillance, citing a lack of consent and concerns regarding data privacy.
• The company is currently investing up to $145 billion this year into AI initiatives amid ongoing internal restructuring and layoffs.
• Meta stated there is no evidence that any employee data was improperly accessed during the two-month duration of the program.

Why it Matters

This incident highlights the growing tension between aggressive corporate AI development and employee privacy expectations within major technology firms. It underscores the significant operational and morale risks companies face when implementing invasive data collection practices to maintain competitiveness against rivals like OpenAI and Anthropic.