Meta’s Chief Technology Officer Andrew Bosworth had a blunt message for employees unhappy about the company’s new keystroke and mouse-tracking software: deal with it. When workers flooded an internal thread asking how to opt out, Bosworth replied plainly that “there is no option to opt out of this on your work provided laptop.” The comment, first reported by Business Insider, landed to a chorus of crying, shocked, and angry-face emojis from colleagues. The angry-face emoji, notably, was also the most common reaction to the original announcement itself.The tool is called Model Capability Initiative, or MCI. Rolled out this week to US-based full-time employees and contingent workers, it logs mouse movements, click locations, keystrokes, and screen content across a pre-approved list of work apps—Gmail, GChat, VSCode, and Meta’s internal AI assistant Metamate among them. The goal, per an internal memo obtained by Business Insider, is to train Meta’s AI agents to handle the mundane computer tasks humans do without thinking—dropdown menus, keyboard shortcuts, navigating interfaces. The stuff that feels trivial to people but still stumps AI models.
Meta’s AI agents need to learn the basics of being human at a desk
“For agents to understand how people actually complete everyday tasks using computers, we need to train our models on real examples,” the internal announcement read. A Meta spokesperson confirmed the program to multiple outlets, adding that “safeguards are in place to protect sensitive content” and that the data won’t be used for performance reviews or anything beyond model training. The tracking does not extend to phones, only work computers.It’s worth noting, per a person familiar with the matter cited by Business Insider, that Meta employees’ activity on work devices has long been monitored—and workers are informed of that when they sign on. In that sense, MCI is more of an extension of existing policy than a brand-new surveillance push. That context, however, has done little to soften the reaction on the ground—especially given what else is happening at the company right now.The MCI rollout landed the same week Meta informed employees it would be cutting 10% of its global workforce—roughly 8,000 jobs—with another 6,000 open roles being closed, according to an internal memo reported by the New York Times. Meta’s chief people officer Janelle Gale framed it as a trade-off to fund other investments, AI chief among them. The company is expected to spend up to $135 billion on AI this year alone, nearly double what it spent in 2024.
Employees call It ‘very dystopian’ as Meta doubles down on its AI obsession
The human cost of that spending is already becoming visible beyond Meta’s own walls. As WIRED reported, more than 700 workers at Covalen—a Dublin-based contractor handling data annotation and content moderation for Meta—were told their jobs are at risk. Around 500 of them are data annotators whose job is to check AI-generated content against Meta’s safety rules, essentially training the models to make better decisions. They were informed over a brief video call on a Monday afternoon and were not allowed to ask questions. Between this round and a previous cut in November, Covalen’s Dublin headcount is on track to be nearly halved, according to the Communications Workers’ Union. “It’s undignified,” one affected employee told WIRED.For Meta’s own employees watching all of this unfold, the MCI announcement hit differently than a standard policy update. One current staffer told the BBC it feels “very dystopian.” A recently departed employee called it “the latest way they’re shoving AI down everyone’s throat.” The top comment on the internal announcement was simply: “This makes me super uncomfortable. How do we opt out?”Now they have their answer. Across the company, Meta has been reorganizing staff into “AI pods,” launching dedicated AI Weeks, and forming Meta Superintelligence Labs. MCI is the latest piece of that machine—and for the people powering it, apparently, participation isn’t optional.
Full internal MCI announcement:
As Mark and Alexandr recently shared, our launch of Muse Spark is the first in a series of new large language models from MSU. We’re on a really strong trajectory with our models and one of the ways we can accelerate our path is by tapping into our own work day to day. While AI models excel at research and technical skills like coding, they still lack some of the basic ways that humans use computers like choosing from dropdowns and keyboard shortcuts. For agents to understand how people actually complete everyday tasks using computers, we need to train our models on real examples.This is where all Meta employees can help our models get better simply by doing their daily work.Starting today, we’re rolling out a tool for US-based FTEs and Contingent Workers that captures computer inputs like mouse movements, click locations and keystrokes as well as screen content for context.Scope is limited to a pre-approved list of work-related applications and URLs, like Gmail, GChat, Metamate, and VSCode. US-based employees will see a pop-up with instructions to enable the tool called Model Capability Initiative (MCI).This only applies to your computer and not your phone. To learn more about how the tool works including privacy safeguards, check out the wiki and FAQs.
Leave a Reply