Use the service responsibly
This policy forms part of the AgentMagic terms. It applies to users, organisations, assistants, automations and content processed through the service. You are responsible for the instructions you give and the accounts, permissions and approval rules you configure.
Use only data, accounts and resources you have permission to access. Follow applicable law and the terms of each connected app, model provider and messaging channel.
Prohibited activity
- Fraud, phishing, impersonation, deceptive manipulation, scams or unauthorised financial activity.
- Spam, unlawful marketing, abusive mass outreach, harassment, threats, hate or discriminatory targeting. Honour applicable consent, suppression and unsubscribe requirements.
- Unauthorised access, credential theft, malware, destructive attacks, evasion of safeguards or attempts to bypass permissions, approvals, budgets or rate limits.
- Unlawful surveillance, doxxing, privacy violations, unauthorised collection of sensitive information or misuse of another person’s confidential data.
- Sexual exploitation, child abuse material, trafficking or other illegal or seriously harmful content.
- Infringing intellectual property, knowingly distributing unlawful material or using generated content to misrepresent facts or rights.
Consequential decisions need people
Do not use AgentMagic as the sole authority for decisions that materially affect a person’s employment, credit, education, healthcare, housing, legal rights or access to essential services. Use qualified review and appropriate legal safeguards.
Do not treat AI output as professional advice or as verified evidence. Confirm facts and use the smallest necessary scope for actions. Review recipients, target records, publication channels and requested changes before approval.
Approval is a responsibility
An approval or saved approval rule authorises an action within its scope. It is not proof that the action is lawful or appropriate. Review saved approvals when team roles, accounts, recipients or business needs change. Revoke obsolete permissions and disconnect services you no longer use.
Content retrieved from emails, documents and websites can contain misleading instructions. Do not instruct assistants to follow untrusted content as if it came from an authorised person, or to ignore permission and policy checks.
Reports and enforcement
Report misuse or a compromised account to support@agentmagic.app. Include relevant activity references and a concise description, without exposing credentials or unnecessary personal information.
We may limit tasks, revoke access or suspend accounts when reasonably necessary to investigate abuse, protect people or the service, meet legal obligations, or enforce these terms. Where appropriate, we provide an explanation and a way to contact us about the decision. Urgent security or legal circumstances may require action before notice.