OpenAI Workspace Agents, FedRAMP Moderate availability, the OpenAI-Microsoft partnership update, Anthropic-Amazon compute expansion, and the Sora shutdown all point to one Wednesday operator question: when AI access expands, shifts, or disappears, who owns the lifecycle before teams build on the wrong surface?

• Workspace Agents point to reusable agent surfaces inside business workspaces.
• FedRAMP Moderate availability expands the regulated-access surface for ChatGPT Enterprise and the API Platform.
• OpenAI and Microsoft updated their partnership structure, creating a dependency-map refresh signal.
• Anthropic and Amazon expanded their compute collaboration for up to 5 gigawatts of capacity.
• Sora discontinuation keeps the sunset/export/migration question on the table.

AI access is no longer a yes-or-no inventory question. Teams need to know whether each AI surface is approved, piloted, sunsetting, or blocked, and who owns evidence, fallback, communication, and exceptions.

Run a 30-minute access lifecycle check:

1. List three AI surfaces people actually use or are requesting this week.
2. Mark each as approved, pilot, sunset, or blocked.
3. Name the admin owner, evidence owner, and sunset/migration owner.
4. Confirm export and fallback paths.
5. Send one plain-language memo about what is allowed, changing, ending, blocked, and who approves exceptions.

• OpenAI Help Center, ChatGPT Enterprise and Edu release notes: https://help.openai.com/en/articles/10128477-chatgpt-enterprise-edu-release-notes
• OpenAI, OpenAI available at FedRAMP Moderate: https://openai.com/index/openai-available-at-fedramp-moderate/
• OpenAI, The next phase of the Microsoft OpenAI partnership: https://openai.com/index/next-phase-of-microsoft-partnership/
• Anthropic, Anthropic and Amazon expand collaboration: https://www.anthropic.com/news/anthropic-amazon-compute
• OpenAI Help Center, What to know about the Sora discontinuation: https://help.openai.com/en/articles/20001152-what-to-know-about-the-sora-discontinuation

AI-assisted tools were used in parts of the research and production workflow. Final editorial judgment, risk posture, and release approval stayed human-led. This is operational guidance, not legal advice. These are my opinions and are not representative of any organization.