Reaching & Rooted Publishing

AI Data Handling & Manuscript Privacy Policy

Last updated September 2, 2026

1. Purpose

This policy explains how Reaching & Rooted Publishing (“R&R”) uses artificial intelligence tools in the course of manuscript development, editing, client calls, and related services, and the specific protections in place to ensure that client (author) content is never used to train third-party AI models and is not exposed to unauthorized parties.

2. Scope

This policy applies to all AI-assisted work performed on client manuscripts, transcripts, and original written or spoken content, including but not limited to developmental editing support, structural analysis, copyediting assistance, manuscript-related workflow automation within Reaching & Rooted's internal systems (“RR-OS”), and AI-assisted meeting recording and transcription used during client calls.

3. Which AI Tools We Use

3.1 Manuscript & Editorial AI (Claude / Anthropic)

R&R uses Claude, an AI model built by Anthropic, accessed exclusively through Anthropic's commercial business channels:

• The Anthropic API, integrated directly into RR-OS, our internal operating system, for manuscript analysis and editing-support workflows.

• Anthropic business accounts (not individual personal/consumer accounts) for any supplementary manual work.

R&R does not use free or personal-tier consumer AI accounts (e.g., an individual staff member's personal ChatGPT or Claude.ai Pro login) for any client manuscript work, under any circumstance.

3.2 Meeting Recording & Transcription (Fathom)

R&R uses Fathom, an AI meeting notetaker, to record, transcribe, and summarize select client calls for internal reference and follow-up. Fathom is a separate product from R&R's manuscript-editing AI tools and is used solely for meeting documentation, not manuscript content.

4. How Client Content Is Protected

4.1 No Third-Party Model Training on Client Content

Anthropic's commercial terms of service — which govern API and business-account usage — contractually exclude customer content from being used to train Anthropic's AI models. This is a standing term of the commercial agreement, not an opt-in setting we configure manually.

Separately, Fathom's privacy policy states that none of its AI sub-processors (including Anthropic, OpenAI, and Google) are contractually authorized to use R&R's meeting data to train their models.

4.2 Limited, Automatic Data Retention (Anthropic)

Content sent through Anthropic's commercial API is automatically deleted from Anthropic's backend systems within a short window (industry-standard practice is 7–30 days), used only transiently to generate the requested output. No standing archive of manuscript content is retained by Anthropic beyond this window.

4.3 Fathom's In-House Model Training — Disabled

Fathom may, by default, use de-identified meeting content to improve its own in-house AI models (separate from the third-party protection above). R&R has this setting reviewed and disabled at the account level so that no R&R meeting content is used for this purpose. Team-level accounts allow this to be disabled organization-wide by an administrator rather than relying on each individual user.

4.4 Controlled, Single-Pipeline Access

All AI-assisted manuscript work is routed through RR-OS's internal pipeline rather than ad hoc use by individual staff members. This means:

• A single, business-owned API credential is used for all manuscript AI work — eliminating the risk of a staff member inadvertently using an unprotected personal account.

• Every AI-generated analysis or edit passes through a human review checkpoint before anything is shared with an author or acted upon.

• Access to manuscript content within RR-OS is limited to designated team members involved in the editorial workflow.

4.5 No Public or Free-Tier Exposure

Client manuscripts are never entered into free, public, or personal-tier AI tools. This eliminates the risk associated with consumer AI products, which may retain conversation history or, in some cases, use content for model training absent an explicit opt-out.

5. What This Means for Authors & Clients

In plain terms: your manuscript, transcript, or original content is processed using AI as an editorial support tool — similar in spirit to spell-check or other editing software — under a business agreement that legally prohibits that content from being used to train the AI itself, and ensures it is not stored indefinitely on any third-party server. If a call with our team is recorded via Fathom for internal reference, the same no-third-party-training protection applies, and R&R has disabled Fathom's own model-improvement use of that content. A human editor at R&R reviews all AI-assisted work before it is incorporated into your project.

6. Ongoing Review

This policy reflects R&R's practices and the published commercial terms of Anthropic and Fathom Video Inc. as of the date above. Vendor policies are subject to change; R&R will review this policy periodically — including rechecking vendor terms after any vendor ownership change — and update it as needed to ensure continued accuracy.

Internal note: this draft is intended as a starting point for review with legal counsel before publishing externally or incorporating into R&R's client-facing privacy policy.

Contact Us

If you have any questions about this policy, contact us by email here:

[email protected]