Meta has officially ended an internal performance policy that required employees to show AI usage and token consumption during reviews, according to internal communications obtained by WIRED. The company updated its performance review guidelines to prioritize actual work impact regardless of whether AI tools were utilized, removing designations like “AI Native” and internal dashboards that tracked token usage.
Simultaneously, Meta has begun internal testing of Hatch, a new agentic AI tool capable of browsing the web and carrying out autonomous actions across desktop applications. Employees have been encouraged to test Hatch on corporate devices ahead of a planned public release.
However, the rollout has raised internal privacy concerns among workers hesitant to connect personal accounts due to fears of data mishandling or operational errors. Meta previously collected employee keystroke data on work devices for AI model training.
Why it matters
Reflects corporate recalibration away from vanity AI adoption metrics toward outcome-focused productivity evaluations.
Highlights internal privacy and reliability challenges when testing agentic desktop AI tools prior to public deployment.
Source: wired.com
