← Back to Workbench

AI Disclaimer

Last Updated: May 2026 · Pre-release notice

Author decision first: Palim is designed around inspectable AI proposals. AI suggestions should remain reviewable material until an author chooses what to apply.

This AI disclaimer explains intended beta boundaries for optional AI-assisted writing capabilities in Palim. Final production language will be reviewed before general availability.

1. Proposals, Not Overwrites

AI Copilot and Revision Workbench are designed to stage suggestions as proposals, not silent manuscript overwrites. Authors should review, edit, accept, or reject suggested changes before treating them as part of the manuscript.

2. Factuality & Hallucination Risk

Large Language Models are probabilistic generators. They can output factually incorrect information, create historical anachronisms, introduce continuity errors, or fail to adhere to physical realities. You, the author, serve as the ultimate editor. We do not guarantee the historical, factual, or logical accuracy of any text suggested by the AI.

The copyright status of AI-assisted text depends on applicable law and the terms of the selected AI provider. Palim's product direction is to avoid claiming ownership over author manuscript materials.

4. Connection Integrity & API Failures

When authors connect third-party or local AI services, availability and output quality depend on those services. Rate limits, downtime, policy changes, and costs should be checked against the provider's current terms.

5. No Automated Submission

Do not send sensitive manuscript context to an external AI service unless you understand that service's data handling policy. Palim should not imply that third-party providers share one universal privacy model.