If we use AI, we say so.

Effective date: 19 August 2026. This is the Signet7 Artificial Intelligence Use Policy.

1. What Signet7 is

Signet7 is a cryptographic email-evidence product and an optional action gate. It is not a generative chatbot. It does not write email, chat replies, testimonials, or marketing copy.

2. What we do not do

• No large-language-model writes Signet7 replies, seals, or claims.
• The mailbox agent uses IMAP, Microsoft Graph, or a local folder. It does not call an LLM.
• Customer content is not used to train a generative model.
• We do not hide AI use. If a later release adds a generative model, this page will name the model class, the purpose, the data sent, and whether customer content is used for training. Training stays off unless a later written term says otherwise.

3. What we do

An optional policy engine can record whether a named action such as EXECUTEWIRE, PURCHASE, REFUND, or TRANSFER was allowed or blocked. That is a static rule check against provenance and authorization state. It is not content generation and not a fraud verdict.

4. Crisis hold

If crisis language appears in a decision, MCP, or mailbox-agent path, Signet7 holds the action and points to 988 instead of issuing a seal or policy verdict. See Safety.

See Privacy Policy.