This policy is part of the Terms of Service. It applies to everyone who uses Steward Agents through your organization's account. If you're unsure whether something is allowed, ask us first: [support email].
1. No client or family health or case information
Do not put personal health information or case information about clients, participants, families or other people your organization serves into Steward Agents. That includes names together with diagnoses, medications, treatment, hospital stays, crisis or safety details, case notes, assessments and intake records, in prompts, selected text, files, emails or transcripts.
- Use de-identified or aggregate information instead, for example "a parent called about discharge planning" or "38 of 44 participants reported more confidence".
- Steward Agents checks text for signs of this information and refuses to send it to the AI. That check is a safeguard, not permission: do not try to get around it.
- If you put this information in by mistake, delete the conversation and tell us at [privacy contact email].
2. Not a clinical or decision-making tool
Do not use Steward Agents to make or support clinical, medical, counselling or safety decisions, or decisions about a person's eligibility for services, funding, employment or housing. Agents may help staff write about programs in general; decisions about individuals stay with qualified people.
3. A person reviews before anything goes out
- Do not send, publish, post, submit or sign agent output as if a person had reviewed and approved it when no one has.
- Do not connect Steward Agents to other tools, scripts or automations so that its drafts are sent or published without a person's review.
- Check facts, figures, names and commitments in every draft before using it, especially in anything going to a funder, board, government, media or the public.
4. Not allowed
- Anything illegal, or that infringes someone else's rights, including copyright, privacy or confidentiality.
- Harassment, threats, hateful or discriminatory content, or content that sexualizes minors.
- Misleading or deceptive material, impersonating a person or organization, or inventing facts, quotes, statistics or endorsements and presenting them as real.
- Sending spam or commercial electronic messages without the consent Canada's anti-spam law (CASL) requires.
- Partisan political activity by registered charities, in breach of CRA rules.
- Trying to get around the service's safeguards, limits or access controls, probing for vulnerabilities without our permission, or disrupting the service.
- Reverse engineering, decompiling or copying the service, except where the law expressly allows it despite this restriction.
- Using Steward Agents to build a competing product, or reselling access without our agreement.
5. Accounts and sign-in
- Each person uses their own Microsoft work account. Do not share sign-ins, let someone else use your session, or use another person's account.
- Keep your Microsoft account secure, using your organization's multi-factor authentication where available.
- Tell your administrator and us at [security contact email] right away if you think your account has been misused.
6. If this policy is breached
We may remove content, suspend a user, or suspend your organization's access if that's needed to stop a serious breach, as described in the Terms of Service. Where we can, we'll tell you first and work with you to fix it.