On September 8, 2026, Microsoft announced that it would review the use of AI by employees and tighten management controls. The company required approximately 350 American employees to report their AI expenditures over the past 28 days. The highest recorded expenditure was $28,000 (approximately 190,000 yuan), and the median amount reported voluntarily was $300 per month. Microsoft has implemented department-level budget control, included individual token expenditures in dashboard monitoring, and switched the internal default models to OpenAI’s GPT-5.6 Sol. In a memo, CEO Jay Parikh criticized excessive data processing, emphasizing that focus should be on creating real value for customers rather than simply increasing data volume. This move marks Microsoft’s official integration of AI expenditures into financial budget management, following the release of Claude Code in December 2025 and the cancellation of most employee permissions in May 2026.