• Claude only: Our enterprise agreement ensures Anthropic does not train on company data — meaning anything you type into Claude is never used to teach or improve Anthropic's AI models, and your inputs are not stored or reviewed for that purpose — the company owns what we submit and receive, and content is access-controlled. These protections apply to our enterprise agreement and do not exist on consumer or personal accounts of any AI tool — including a personal Claude account. Claude is the sole approved standalone AI tool for company work: do not use ChatGPT, Gemini, Perplexity, Grok, or comparable services. AI built into Spire-licensed products — such as Copilot in O365 or Adobe AI Tools — is authorized.
• Business use only: Claude is provided as a business tool. Personal projects, outside ventures, and assistance to third parties are not permitted.
• Accountability rests with you: Claude may produce errors. Review, verify, and apply your professional judgment before any output is relied upon or distributed.
• Company ownership: Anything you submit to or create with Claude belongs to Spire Hospitality. This applies to all work product, regardless of topic or context.

• Think before you paste: Publicly available information, your own drafts, and general internal material with no sensitive details are fine. Stop and get approval if it includes personal data, credentials, or confidential business information. Privileged matters cannot be entered into Claude at all, with or without approval.
• Protect proprietary information: Trade secrets, unreleased plans, and sensitive financial data require prior written approval.
• Exclude personal data: Names, addresses, SSNs, account or payment details, wire instructions, and health information may not be entered unless an approved protocol permits it. This includes guest data of any kind — reservation records, loyalty profiles, folios, and payment card data. Payment card data is also subject to PCI-DSS and must never be entered into Claude under any circumstance.
• Safeguard credentials: Passwords, API keys, and access tokens must never be entered into Claude.
• Know what Claude retains: Claude can carry forward context from prior conversations — your role, ongoing priorities, and details from past chats. Review and manage what's retained through Settings → Memory.
• When in doubt, ask first: Route employee matters to the SVP of People & Culture; route all other matters to the SVP, Owner Services.

• Preserve human judgment: Claude must not make final determinations on hiring, compensation, legal, or financial matters.
• Review for bias and accuracy: Examine outputs for fairness and errors, particularly any material affecting people, owners, or guests.
• Disclose where required: Indicate AI assistance whenever a client contract, regulation, or specific assignment calls for it. For external-facing deliverables, disclose where AI was materially involved in the substantive content — analysis, recommendations, or original drafting — even if no contract specifically requires it. Editing, formatting, and polishing your own work do not require disclosure.
• Report concerns immediately: Escalate errors, data exposure, or any irregularity to your manager or the Owner Services team without delay — speed matters more than self-diagnosis. For suspected data exposure, contact by phone or in-person rather than email, to avoid any delay.

Use of Claude may be logged, monitored, and audited. Access is allocated to users who actively engage with the pilot and apply the training; it may be modified, suspended, or revoked at the company's discretion. Adherence to this Code is a condition of continued Claude access; concerns or violations should be reported to your manager or the Owner Services team.