Use any AI.
Keep every privilege.
Privil sits between your firm and every consumer AI, ChatGPT, Claude, Gemini, Copilot.
Privileged content stays inside the firm. Discoverable content never leaves.
Built for firms of 50 to 500 attorneys.
After Heppner,
every prompt is on the record.
United States v. Heppner, S.D.N.Y., February 17, 2026
In a unanimous opinion, Judge Rakoff held that conversations between attorneys and consumer-facing AI systems, ChatGPT, Claude, Gemini, Copilot, are notprotected by the attorney–client privilege.
The court reasoned that a third-party AI provider is not a lawyer, not an agent of a lawyer, and not retained for the purpose of providing legal services. Lawyers transmitting client confidences into a commercial AI system, the court found, do so with no reasonable expectation of confidentiality.
The opinion is binding on the Southern District of New York and has already been cited approvingly in the Second, Third, and Ninth Circuits. State bar associations in California, New York, Florida, and the District of Columbia have issued conforming ethics guidance.
Every prompt your lawyers have ever sent to ChatGPT or Claude is potentially discoverable.
Privil makes sure the next one isn’t.
79%
of lawyers use AI in their practice.
ABA Tech Report, 2025.
10%
of firms have a formal AI policy.
Thomson Reuters Future of Professionals, 2025.
A written policy is a sticky note on an unlocked door.
Privil enforces your policy at the prompt level, before sensitive content ever reaches a third-party AI provider.
Three steps. Zero change to how your lawyers work.
Lawyer prompts an AI tool.
They use Claude, ChatGPT, Gemini, or Copilot exactly as they do today. No new interface. No retraining.
Privil inspects the prompt.
Legal-aware detection identifies privileged content using your firm's matter taxonomy, client list, deal codes, and ABA-aligned rules.
Policy applies.
Block, redact, or route to enterprise-tier AI. Every action is logged to a defensible audit trail.
Logged · Sealed · No. 24-pr-00115
Three deployment models. Pick what fits your firm.
API Proxy
One line of configuration.
We sit in front of every API call your firm tools make to Anthropic or OpenAI.
- Anthropic
- OpenAI
- Gemini
- Bedrock
Browser Extension
For chatgpt.com, claude.ai, and the rest.
Deployed via Google Workspace or Intune. Catches prompts pasted into chatgpt.com, claude.ai, and other web interfaces before they're submitted.
- Google Workspace
- Microsoft Intune
- Jamf
Network Gateway
For mature security stacks.
Integrates with Zscaler, Netskope, and Cloudflare for Teams. Bulletproof for firms with mature security postures.
- Zscaler
- Netskope
- Cloudflare
Built for the practice of law.
Not adapted from generic DLP.
Privilege-aware detection.
Recognizes client names, matter numbers, deal codes, case captions, opposing parties, and litigation patterns, not just generic PII.
Heppner-aware audit trail.
Every AI interaction logged and mapped to ABA Model Rule 1.6, with exports formatted for state bar reporting.
Document management integrations.
Pulls live taxonomies from iManage and NetDocuments so detection reflects the firm's actual matter universe.
Tool-agnostic.
Works with Claude, ChatGPT, Gemini, Copilot, and any AI tool your lawyers adopt next.
Why nothing else does this.
| Capability | PrivilPrivilege-aware AI privacy | Legal AI AppsHarvey, CoCounsel | AI GatewaysLiteLLM, Portkey | Legacy DLPSymantec, Forcepoint |
|---|---|---|---|---|
| Built for law firms | ||||
| Privilege-aware detection | ||||
| Works with existing AI tools | ||||
| Bar-compliance audit trail | ||||
| iManage / NetDocuments integration | ||||
| Browser-level enforcement |
For mid-market firms with active AI usage and post-Heppner anxiety.
Privil is sized for firms of 50 to 500 attorneys, large enough that AI is already in use, small enough that policy enforcement is still a fight you can win.
Priced like enterprise software.
Quoted like outside counsel.
Pricing scales with firm size and AI usage volume.
Contact us for a quote tailored to your practice mix and deployment needs.
- ▸Per-attorney annual licensing
- ▸Volume-based AI usage tiers
- ▸Optional on-premise deployment
- ▸Annual contracts, with NDA upfront