Meta to Use Employees' Click Patterns for AI Training [Global AI Briefing]

[Financial News] Meta, which has recently been pouring resources into strengthening its artificial intelligence competitiveness, has introduced software that tracks employees' mouse clicks and keyboard inputs and plans to use the data for AI training. The move has sparked strong backlash inside the company, with critics calling it corporate surveillance.
Meanwhile, overseas markets such as India have recently seen a growing push to collect and use real-world work patterns as AI training data. In some call centers and sewing companies, employees are being monitored with cameras and sensors during working hours, and their entire work processes are being recorded for AI learning.

As competition among AI models intensifies and the use of AI expands across industries, big tech companies are increasingly trying to use real human work patterns in key sectors as training data.
According to major foreign media outlets including Business Insider, Meta installed software on employees' computers that can collect information such as mouse movements, button clicks, dropdown menu navigation and keyboard input.
Meta said the data will be used to learn how people perform tasks in real work environments, with the goal of improving AI productivity and accuracy.
However, the move has met with strong resistance among employees. Some workers see it as surveillance and have raised concerns that their behavior at work is being continuously recorded.
In response, Meta explained, "This system is not intended to monitor individual employees, but to collect data for improving AI models," and added, "The collected data will be used with appropriate safeguards, and personally identifiable information will be minimized."


Anthropic spent $1.6 million, or about 2.3 billion won, on government lobbying in the first quarter of this year, a fourfold increase from $360,000 in the same period last year. OpenAI also spent $1 million on lobbying, suggesting that AI companies are stepping up their efforts to influence the government.

According to Axios on the 22nd local time, Anthropic spent $1.6 million and OpenAI spent $1 million on lobbying in the first quarter, marking record-high lobbying expenditures. Compared with the first quarter of last year, when Anthropic spent $360,000 and OpenAI spent $560,000, lobbying costs have risen sharply.
Anthropic provided its AI model Claude to the United States Department of Defense (DoD), but tensions arose as the company maintained its principle that AI should not be used for mass surveillance in the United States or for autonomous lethal weapons without human oversight. The executive branch of the United States federal government later designated Anthropic as a "supply chain risk" company, and the resulting disputes appear to have driven broader lobbying efforts to resolve the issue. OpenAI said it has lobbied on issues related to AI, copyright, cybersecurity, cloud computing and infrastructure.

cafe9@fnnews.com Lee Gu-soon Reporter