Meta has formally ended a program that tied employee performance reviews to how much they used artificial intelligence tools, according to an internal announcement reported this week by WIRED. Three workers who received the message confirmed the social media company will no longer evaluate staff based on their AI consumption. At the same time, employees are testing a new autonomous AI agent called Hatch that can independently perform tasks on a computer, and the tool is driving token usage even higher than before.
Nearly a year ago, Meta told workers they'd be assessed on their "AI-driven impact," which meant gauging how extensively they deployed chatbots and agents to enhance daily tasks. This approach led to employees receiving classifications like "AI Native," "AI First," or "AI Enabled" based on consumption levels, according to a July lawsuit filed by roughly two dozen workers who alleged Meta violated antidiscrimination laws during May layoffs. The new guidance unveiled this week removes language about evaluating staff based on "usage of AI" and their "AI Native" status, replacing it with more flexible phrasing that notes "these outcomes can be supported by AI or other means." Engineers across the company were informed that Meta "will not use AI adoption dashboards or token counts to evaluate impact," The Information reported Wednesday. Some employees described the shift as subtle but meaningful, eliminating pressure to deploy AI in situations where it doesn't make sense.
For months, workers say, colleagues prompted AI tools repeatedly—or what they considered frivolously—to inflate internal consumption metrics like token counts. The tokenmaxxing period peaked when one employee built an internal scoreboard tracking how much staff used AI, assigning labels such as "Token Legend." After information about the dashboard leaked, the worker removed it in April. A few months later, Meta began rationing employees' AI access. Meta spokesperson Tracy Clayton said the updates aim to emphasize what has always been true—that the company assesses workers based on their contributions—and that designations like "AI Native" were never employed for performance evaluation.
More recently, Meta has been encouraging—though not mandating—staff to use its most sophisticated AI experiment, an agentic tool capable of autonomously executing actions on a computer. Known as Hatch and comparable to the viral sensation OpenClaw, it can navigate the web and operate other applications. Workers have been able to test Hatch for several weeks on corporate devices ahead of an anticipated public launch. Some employees have hesitated to connect Hatch to personal email, calendar, and other non-work digital accounts due to privacy concerns and the risk that the AI could damage something critical, two current workers familiar with the worries told WIRED. Meta previously ran a project that monitored employees' keystrokes and other activity on work devices to gather data for training AI systems, a program the company has since paused that eroded many workers' confidence in its AI tools. However, no longer fearing penalties on performance reviews for avoiding AI testing could help lift morale and ultimately rebuild faith among Meta's workforce. Already, some staff say they've adopted Hatch for booking personal appointments and managing life outside work, describing the tool as effective though some worry about the environmental and economic costs of consuming large volumes of tokens and computing resources to automate tasks.
Though workers may no longer be deliberately maximizing token usage, Hatch is still consuming more than standard chatbots and coding assistants, employees say. And staff believe management remains interested in seeing them demonstrate they can effectively wield AI tools. If Hatch ends up increasing productivity as intended, some workers fear Meta could execute additional job cuts. Last week, Reuters reported Meta was preparing for another round of major layoffs on top of the 8,000 positions eliminated in May, but called off the plans amid a rise in software bugs linked to dependence on AI and slower-than-expected progress on agent development. CEO Mark Zuckerberg has said further mass layoffs aren't expected this year. The policy shift signals Meta is recalibrating how it balances AI adoption with employee autonomy, attempting to foster innovation without mandating it through performance metrics. Removing explicit quotas may reduce gaming behavior while still pushing the frontier on agentic tools that could reshape how work gets done—or how many people are needed to do it.