Microsoft Cracks Down on Excessive AI Token Usage Among Employees

Microsoft has been forced to introduce new guidelines to curb excessive AI token usage among its employees. The move is aimed at cutting down on 'tokenmaxxing', where staff use more tokens than necessary, and instead focus on maximizing outcomes that benefit customers and the business.

The company's executive vice president Jay Parikh wrote in an email to employees that they should not be optimizing for fewer tokens, but rather for greater value from their token investment. He noted that many engineers spend hundreds or thousands of dollars a month on AI tokens, and that Microsoft wants to ensure it is getting the most out of its AI platforms.

As part of these new guidelines, Microsoft will make access to the cheaper OpenAI GPT-5.6 model the default for internal use. Employees will also be able to track their individual AI spending, with an 'AI token budget target' set for each division. The company aims to achieve greater value from its token investment without impairing progress towards becoming 'AI-first', and will continue to adjust its policies as models and products evolve.

Microsoft's decision comes after reports of excessive AI usage among employees, with some using tools like GitHub Copilot CLI and Claude Code to the point where they exceeded their annual budget in just four months. The company has previously cancelled most of its Claude Code licenses and told engineers to use GitHub Copilot CLI instead.