Meta's AI Usage Leaderboard Taken Down After One Week, Employee Quota-Filling Scheme Exposed

According to 1M AI News monitoring, the AI Usage Leaderboard within Meta known as Claudeonomics has been taken down approximately a week after going live. The Information reported that the creator voluntarily removed the leaderboard, not due to a request from Meta management. Prior to its takedown, the leaderboard displayed that Meta's collective 30-day token consumption had risen from 60.2 trillion a week ago to 73.7 trillion, with the top individual user consuming 3.285 trillion tokens, roughly equivalent to nearly $2 million at public pricing.

The leaderboard sparked internal controversy. An employee wrote on an internal forum: "I suggest everyone roughly estimate the energy consumption behind this. If it weren't so sad, it would be a joke." Meanwhile, various forms of token gaming came to light: some engineers had AI agents generate a large number of meaningless tiny code commits, with each commit counting as token consumption; other employees developed a transcription bot to passively record and take notes during meetings, with the bot's token consumption attributed to the creator. Similar behavior had also been seen internally at Amazon, where a team scripted conversations to make it appear that each interaction with the AI programming tool Cline consumed ten times the tokens, propelling the team to the top of their department's usage leaderboard until Amazon patched the loophole earlier this year.

While the leaderboard has been removed, Meta continues to track employee and team token usage through the official AI Insights dashboard, and the performance assessment system, Checkpoint, includes token consumption as one of the evaluation metrics. Some low-usage employees have privately expressed concerns, fearing they may not be seen as "AI native."