AI Use and Disclosure

This website has no chatbot. Forms are read by a person. Automated security checks validate a submission and reduce abuse. They do not decide the scope or acceptance of an engagement.

For new engagements, our default provider is the OpenAI commercial API. The Anthropic commercial API is an alternative. A self-hosted model requires a separately agreed environment and data handling plan. The engagement identifies the tools, vendors, processing locations, permitted data, and responsibilities before implementation.

Our policy requires human review of output before delivery, including factual checks and review for unintended disclosure. The responsible client approves consequential actions. We document corrections and review recurring errors.

OpenAI's published API terms exclude model training by default. Abuse-monitoring content is ordinarily retained for up to 30 days, with exceptions. Features can retain additional application data.

Anthropic's published commercial terms exclude model training by default. Its API retention policy ordinarily deletes inputs and outputs within 30 days, with exceptions. Saved features and safety requirements can retain data longer.

Our delivery policy prohibits model-training opt-ins. Confidential client information requires an approved commercial service and agreed data handling terms. Provider selection does not establish zero retention.

AI output can contain mistakes. The agreed workflow must identify who reviews output and how consequential decisions are checked.

Some website copy is drafted with AI assistance and reviewed by Oscar G.

Arx Sable does not provide licensed psychological services. Its AI tools and implementation work are not a substitute for licensed medical or psychological care.

Arx Sable does not currently offer a publicly accessible generative AI system of its own.

Arx Sable does not use automated decision-making to make legal or similarly significant decisions about site visitors. An inquiry is reviewed by a person.

A future chatbot, public model, or product that processes employee information requires its own review and updated disclosures before launch.

Contact [email protected] about AI use and disclosure.