SDNY Judge Rakoff ruled in January 2026 that AI-generated documents are NOT protected by attorney-client privilege when created with consumer AI tools. Legal Security Guardrails is the compliance layer that changes that.
Using Claude, ChatGPT, or Gemini without enterprise controls constitutes third-party disclosure. Heppner v. US established this risk is real and immediate.
ABA Formal Opinion 512 (2024) requires informed client consent before AI processes privileged materials. Most firms lack documented consent workflows.
When privilege is challenged, you need a documented chain of custody. Consumer AI tools provide none.
Legal Security Guardrails sits on top of Anthropic's Claude Cowork plugin layer. It adds mandatory classification, privilege protection, PII detection, and audit logging — without changing how attorneys use Claude.
| Feature | Free · 10 Uses | Pro · Unlimited |
|---|---|---|
| PUBLIC / INTERNAL classification | ✅ | ✅ |
| CONFIDENTIAL / PRIVILEGED classification | ❌ | ✅ |
| Basic PII detection (SSN, CC, email, phone) | ✅ | ✅ |
| Advanced PII (health, immigration, minor's data) | ❌ | ✅ |
| PII redaction workflow | ❌ | ✅ |
| Audit log (view) | ✅ | ✅ |
| Audit log (export MD / JSON / CSV) | ❌ | ✅ |
| Attorney-client privilege protection | ❌ | ✅ |
| Privilege stamping on outputs | ❌ | ✅ |
| Client consent management (ABA 512) | ❌ | ✅ |
| Ethical wall enforcement | ❌ | ✅ |
| Conflict-of-interest detection | ❌ | ✅ |
| /secure-review command | ❌ | ✅ |
| /compliance-check command | ❌ | ✅ |
| /consent-form command | ❌ | ✅ |
| Security auditor agent | ❌ | ✅ |
| Usage limit | 10 lifetime uses | Unlimited |
Geraci LLP implemented Legal Security Guardrails to protect attorney-client privilege on AI-assisted loan document review and regulatory compliance analysis. With a growing volume of AI-processed client files, the firm needed documented privilege controls and an auditable consent workflow.
"In real estate and private lending law, we handle some of the most sensitive financial and personal information our clients have — loan documents, purchase agreements, and entity structures. At a firm like ours, we don't have the luxury of a massive IT department vetting every piece of technology we bring in. That's why the Guardrails plug-in has been a game-changer. It lets our attorneys harness the speed and analytical power of Claude Cowork without ever exposing confidential client data to outside risk. For any firm looking to modernize without sacrificing the trust their clients place in them, this is the solution." — Anthony Geraci, Founder & Managing Partner, Geraci LLP
Add the plugin via Settings → Plugins → GitHub. Free tier available immediately. Pro access granted after license confirmation.
Every document gets classified (PUBLIC / INTERNAL / CONFIDENTIAL / PRIVILEGED) before any AI processing begins. Client consent and conflict checks run automatically.
Privilege stamping, PII detection, ethical walls, and audit logging run on every session. Export your audit trail at any time.
Legal Security Guardrails addresses the ABA rules and regulations most relevant to AI-assisted legal practice.
No. Legal Security Guardrails is a plugin for Claude Cowork (claude.ai/teams or enterprise). It requires an active Claude Cowork subscription.
Yes. Pro is delivered via a private GitHub repository. Access is granted per licensed seat. No public access exists.
No. Legal Security Guardrails implements the technical safeguards that support a privilege claim. Privilege is ultimately a legal determination. We recommend reviewing deployment with your ethics counsel.
30-day refund from license activation date. Contact info@kaizenailab.com.