Meta expands AI while tracking staff
- Meta rolled out its Meta AI business assistant across major global markets while reportedly installing software to track employee keystrokes and mouse movements. - Reports say the tracking captures activity on sites like Google, LinkedIn and Wikipedia and offered workers no opt-out. - The combination of commercial expansion and internal data collection raises privacy, labour, and legal questions about how firms train enterprise AI. (tech.yahoo.com) (cnbc.com)
Meta is expanding its Meta AI assistant to advertisers worldwide as reports say the company is also tracking employees’ computer activity for AI training. (tech.yahoo.com) (cnbc.com) Yahoo Tech reported on April 23 that Meta widened access to its business assistant across the United States, Europe, the Middle East and Africa, Asia-Pacific, and Latin America, with local-language support for advertisers and agencies. Meta had said in a January 2026 post that it began testing the tool in the fourth quarter of 2025 and planned a broader rollout in the coming months. (tech.yahoo.com) (about.fb.com) The assistant is designed to help businesses with campaign optimization, account support, and personalized recommendations based on a company’s goals, according to Meta’s January announcement. Meta has been pushing more artificial intelligence features into advertising tools as it tries to automate more of the work of buying and tuning ads on its platforms. (about.fb.com) (tech.yahoo.com) At the same time, CNBC reported on April 22 that Meta planned to capture employee keystrokes and mouse clicks on work computers through a program called the Model Capability Initiative. CNBC said internal documents showed the system covered hundreds of websites and apps, including Google, LinkedIn, and Wikipedia. (cnbc.com) CNBC said the monitoring was tied to training artificial intelligence models on how workers complete tasks across websites and software, and that employees were not offered an opt-out. Reuters, cited by CNBC, first reported that Meta had introduced the new tracking tool. (cnbc.com) The two developments land as Meta is trying to sell businesses on AI systems that can remember goals, answer support questions, and suggest actions inside ad accounts. They also show how companies building those systems are looking for large volumes of real workplace behavior, not just public internet text, to improve how assistants perform. (about.fb.com) (cnbc.com) Meta has been widening its AI push well beyond chatbots. In April, the company said its Muse Spark model would power a faster Meta AI assistant across WhatsApp, Instagram, Facebook, Messenger, and its AI glasses, and would also be offered in private preview through an application programming interface to select partners. (about.fb.com) The employee-tracking report adds a labor and privacy dimension to that expansion. Monitoring software that logs clicks, typing, and app usage can show how people navigate tasks step by step, but it can also capture sensitive work patterns and browsing behavior on internal machines. (cnbc.com) Meta’s public materials on the business assistant emphasize convenience for advertisers: faster setup, better optimization, and recommendations tailored to a company’s objectives. The internal monitoring report points to a separate question for workers inside the company: how much of their day-to-day activity can be turned into training data as Meta builds the next version of those tools. (about.fb.com) (cnbc.com)