Meta ends Tokenmaxxing, refocuses employee evaluations on impact
Meta has officially ended its internal 'tokenmaxxing' program, which encouraged employees to maximize AI token usage, often without clear productivity gains. According to a Wired report, Meta communicated this change internally in late August 2026. The company's revised performance review guidelines remove criteria such as 'AI Native' and 'AI First' designations, replacing them with a softer statement that work results 'can be supported by AI or other means.' This follows a lawsuit filed by former employees in July 2026, who claimed unfair treatment due to AI-based evaluations. Meta is also experimenting with a new AI agent tool called Hatch, which has increased token consumption internally. A Meta spokesperson denied that AI usage ever factored into performance reviews.
- •Meta ended its tokenmaxxing program internally in late August 2026.
- •Meta removed AI usage criteria like 'AI Native' and 'AI First' from performance reviews.
- •Former employees filed a lawsuit against Meta in July 2026 over alleged unfair AI-based evaluations.
