Microsoft is rolling out new rules to manage rising AI costs, giving employees dashboards to track spending on AI tokens and setting per-division AI token budget targets. Executive Vice President Jay Parikh explained the changes in an email, aiming for more transparency and control. Parikh said Microsoft isn't trying to cut back on AI, but wants "more impact per token." Although token prices have dropped a lot since late 2022, overall costs have tripled because engineers are accelerating their use of GitHub Copilot. To help, Microsoft set OpenAI's cheaper GPT-5.6 model as the default for internal projects and might add more limits if needed, just like Amazon and Meta are doing to handle their own AI expenses.