AI and Automation Terms

1. Authority and intended use

Use AI and automation only for lawful purposes and systems, data and actions you are authorized to manage. You remain responsible for appropriate notices, permissions and instructions from clients, workers and other affected people. An AI recommendation does not grant authority to perform an action.

Set permissions, scope, approval rules and limits appropriate to the available feature and the impact of a mistake. Do not delegate a decision solely because a tool can technically perform it. Confirm the applicable feature’s supported environments and safeguards before using it in production.

2. Human review and consequential actions

AI output can be inaccurate, incomplete, insecure, biased or unsuitable, even when presented confidently. Review recommendations, messages, scripts, configurations and proposed changes before relying on them. Test consequential changes in a bounded environment where practical and verify the result.

• Obtain approval from an appropriately authorized person before remote control, script execution, destructive infrastructure changes, restoration or deletion, cloud administration, or disclosure or export of secrets.
• Define the targets, expected effect, failure response and a practical rollback or recovery approach before authorizing automation.
• Stop or restrict a workflow when its scope, authorization or outcome is unclear. Do not bypass approval, tenant boundaries, authentication, legal holds or safety controls.
• Do not use these features as the sole decision-maker for legal, medical, employment, credit or other decisions with significant consequences for individuals, or as an emergency or life-safety system.

3. Inputs, providers and charges

Provide only information needed for the task and lawfully permitted for the relevant provider. Do not assume secrets or personal information are automatically masked before transmission. Customer-selected providers may have separate terms, retention, training, security and processing-location practices. Review those before connecting an account or sending data.

You are responsible for the rights and licenses needed for your inputs. Our service agreement does not transfer ownership of your data. Rights in outputs may depend on applicable law and the selected provider; output may not be unique or protectable.

Where you bring an AI-provider account, its fees are paid to that provider unless your order explicitly states otherwise. Availability, rate limits and model changes can affect the workflow. This policy does not promise that a provider retains no data or never uses inputs for model development.

4. Prohibited uses

You must not use AI or automation for malware, credential theft, unauthorized surveillance or access, deceptive impersonation, unlawful discrimination, prohibited content, evasion of trade restrictions, or circumvention of security and privacy controls. Legitimate security work must stay within the owner’s written authorization and the Acceptable Use Policy.

Do not represent generated output as verified product evidence, a certification, professional advice or a human approval when it is not. Preserve required records of authorization and review for your own compliance responsibilities.

5. Problems, changes and remedies

Report suspected harmful output, unauthorized actions or exposed information through the relevant support route; use the legal contact for sensitive concerns and agree a secure evidence exchange. We may restrict a feature where reasonably needed to address a legal or safety risk, subject to the service agreement.

Feature or model changes do not remove expressly agreed material commitments or mandatory consumer rights. Liability, cancellation and dispute provisions follow the accepted Terms; nothing here excuses fraud, unlawful processing or a duty that law does not permit us to exclude.

• Acceptable Use Policy
• Data Processing Addendum
• Support Policy