Meta Ends AI Usage Evaluation Program While Rolling Out New Agentic Tool
Meta has discontinued performance metrics tied to employee AI tool adoption, even as workers begin testing a new autonomous agent called Hatch that continues to drive high token consumption.
What happened
Meta announced this week that it is removing AI usage metrics from employee performance evaluations, abandoning a program that previously graded workers on 'AI-driven impact' and assigned labels such as 'AI Native' and 'AI First.' The change replaces references to 'usage of AI' with looser language emphasizing impact 'supported by AI or other means.' According to The Information, engineers were told the company will not use 'AI adoption dashboards or token counts to evaluate impact.' Simultaneously, Meta employees are testing Hatch, an agentic AI tool capable of autonomous computer actions including web browsing and operating applications. According to employees, token consumption remains high despite the policy shift, and some have hesitated to connect Hatch to personal accounts due to privacy concerns and past company surveillance practices. Some employees have embraced Hatch for booking personal appointments and organizing their life outside of work.
Context
The policy reversal addresses longstanding employee grievances. For months, employees engaged in 'tokenmaxxing'—repeatedly prompting AI tools to maximize internal usage metrics—which reached a peak when an internal leaderboard labeled high-usage workers as 'Token Legend.' A pending lawsuit filed by about two dozen employees alleges Meta violated antidiscrimination laws by laying off workers in May based partly on AI usage scores, with particular harm to those on health and family leaves who could not accumulate usage. Removing usage-based evaluation criteria removes a stated basis for performance penalties and may rebuild trust, particularly given that Meta previously paused a keystroke-tracking project used to train AI systems. However, the introduction of Hatch and continued high token consumption suggests management interest in AI adoption persists through other means. Reuters reported last week that Meta had planned additional major layoffs but called them off due to increased software bugs from AI reliance and slower-than-anticipated agent development progress; Zuckerberg has stated no further mass layoffs are expected this year.
What's disputed
Meta spokesperson Tracy Clayton stated that labels such as 'AI Native' were never used for performance evaluation, contradicting the reporting from Business Insider and the lawsuit description. The degree to which token consumption will continue to rise with Hatch adoption remains unclear, as employees report it consumes more resources than prior tools but Meta declined to comment on testing.