A privileged chat workspace for the people who need counsel.
Privilege AI is the privileged chat workspace for executives, managers, HR, and anyone who should have access to the legal team: open a properly documented privileged AI session, work the question with counsel and the best models, and support counsel-directed AI work with a documented assessment of applicable protections. Authorized and managed by attorneys.

Claude Fable 5.1 · Claude Opus 5.5 · ChatGPT Astra 6

The session is the unit of privilege.
Every conversation belongs to a legal matter. Every answer is recorded with its citations, its model, and who was present — the record that lets counsel support a claim of privilege.
Requesters ask. Counsel reviews the basis and approves.
Three roles — requester, attorney or lead counsel, viewer. Counsel is present in the session, marks the privilege basis reviewed, and decides what leaves the matter.
Citations into matter documents, the model, and who was in the room.
Answers cite the MSA, the vendor notice, the SOC timeline — by section and hash. Each turn is logged to a hash-chained audit ledger with retention and legal hold.
Claude and ChatGPT on privilege-safe routing, with your own keys.
Bring your company's own OpenAI and Anthropic keys, or use a system key. No matter is used to train any model. Web search is off by default and labelled when it leaves the workspace.





Privacy and privilege assertion on a chat and work-product basis.
Provisioned by legal. Used by everyone who asks.
Who opens a session
The people who should have access to legal — and the counsel who authorize them.
Document the legal workflow while the work happens.
Heppner highlights the limits of self-directed AI work followed by legal review after the fact. Privilege AI is designed around counsel’s role from the start: a legal matter, defined participants, a reviewed basis, and a record of the model and cited sources. Give counsel the context needed to direct AI-assisted work and assess the protections that may apply.
Detections from Claude, ChatGPT, and the Privilege AI Agent in one queue. The questions that need legal protection are routed into a Privilege AI session.
On the workstation: 120+ categories of compliance, legal, and regulatory violations, alerted to legal and compliance and recorded on the matter.