Meta Unveils Content From Its Work Series

Meta left critical information collected from employee laptops accessible to anyone at the company, according to an internal security source told WIRED by three employees familiar with the matter.

The data, which was collected as a cross section teach examples of artificial intelligenceis believed to include keystrokes, mouse clicks, and computer displays of US Meta employees.

Meta spokeswoman Tracy Clayton confirms the company is investigating the security issue. “We’ve built the app carefully with privacy protections in mind,” he says, adding, “there’s no indication that anything Meta employees have improperly accessed.”

A security alert posted on Monday revealed that “personal information about 45,000 hive tables,” was exposed. The tables also included staff activities such as “information and documents, private conversations, people and activities,” according to documents seen by WIRED.

Some Meta employees quickly seized on the security failure, saying in internal forums that it confirmed concerns raised when the company began tracking the company’s laptops in April as part of a program called the Model Capability Initiative.

Comments posted on internal forums Monday included questions about how Meta’s privacy checks failed to prevent the breach, and whether anyone whose data was exposed would be allowed to attend the meeting that went wrong, according to documents seen by WIRED.

In one indoor forum where employees are known to sell jokes, an employee posted a meme from Office Jim Halpert holding a sign that read “0 days since our last bullshit.”

Sources at Meta, who were not authorized to speak publicly, told WIRED that the process has now been marked as closed, meaning it may have been terminated. In an internal letter to staff on Monday, Andrew Bosworth, Meta’s chief technology officer, said the implementation of the project did not match what was described in the confidential review and that the findings would be shared.

Last month, more than 1,600 employees at the technology giant signed an internal petition objecting to the evaluation of the laptop, warning that “the collection of this information poses risks to the security and control of Meta, including the possibility of violations and unauthorized disclosure.” The plaintiffs also expressed concern about what they see as the lack of security measures that Meta has implemented. Another expert wrote a widely distributed internal document saying that having their laptop screen for research to study data without their consent was seen as an invasion of privacy and just took advantage of them.

Meta officials previously defended the data collection process, saying it was necessary to train AI systems to use computer programs in the same way humans do. In the audio of the company meeting was downloaded Last month, Mark Zuckerberg, CEO of Meta, told employees that “AI models learn to watch really intelligent people do things” and “the average intelligence of the people at this company is much higher” than ordinary contractors who would have been hired specifically to create this type of data.

But after mounting protests from employees, Meta this month began offering more screening options, including allowing employees to turn off screening briefly to complete critical tasks, such as scheduling appointments, according to two people familiar with the matter. Some workers still want the tracking to be stopped entirely.