ETHICS & COMPLIANCE
Responsible AI
Zuben is built for a profession governed by the highest ethical standards. Our approach to AI reflects those standards at every layer — from architecture to output.
Important Disclaimer
Zuben is a tool designed to assist lawyers in their work. It does not replace the lawyer's independent professional judgment, ethical obligations, or duty to verify all work product. Lawyers remain fully responsible for reviewing, validating, and standing behind all output generated with AI assistance, including output from Zuben. The information on this page is provided for educational purposes and does not constitute legal advice on ethical compliance.
Our Commitment
Legal AI carries unique ethical obligations. Unlike general-purpose AI tools, platforms used in legal practice must comply with the ABA Model Rules of Professional Conduct and the ethical frameworks that govern every aspect of a lawyer's work. Zuben was designed from the ground up with these obligations as architectural requirements — not afterthoughts.
We believe that AI in legal practice must be verifiable, auditable, and transparent. If a system cannot explain how it reached a conclusion, cite the authorities it relied upon, and demonstrate that those authorities actually support its propositions, it has no place in high-stakes legal work.
ABA Formal Opinion 512: The Framework
On July 29, 2024, the ABA Standing Committee on Ethics and Professional Responsibility released Formal Opinion 512, the first comprehensive ethical guidance on lawyers' use of generative AI. This opinion establishes that existing Model Rules fully apply to AI use and identifies six primary areas of ethical obligation. Below we summarize each area and explain how Zuben's architecture addresses it.
Lawyers Must Understand AI's Capabilities and Limitations
Formal Opinion 512 requires lawyers to exercise the "legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation" and to understand "the benefits and risks associated" with AI technologies they use. Lawyers need not become AI experts but must have a reasonable understanding of capabilities and limitations.
Zuben's Approach
Zuben is designed to assist lawyers in fulfilling their competence obligations — not to replace the lawyer's independent professional judgment. The platform's verification gates surface limitations explicitly: when the pipeline cannot verify a proposition, it flags it; when it cannot meet verification standards, it blocks output entirely. These architectural safeguards reduce the risk of undetected errors, but lawyers retain full responsibility for reviewing, validating, and exercising independent judgment over all AI-assisted work product.
Lawyers Must Communicate AI Use to Clients
The opinion provides guidance on when and to what extent lawyers must disclose their use of generative AI to clients. Transparency about the tools used in a client's matter is essential to maintaining trust and informed consent.
Zuben's Approach
Zuben provides tools to support disclosure obligations. Every work product includes a watermark identifying it as AI-assisted output, the pipeline run ID, matter identifier, and classification level. Complete audit trails enable lawyers to explain how AI was used in any matter. However, the decision of what, when, and how to disclose AI usage to clients remains the lawyer's professional responsibility in accordance with applicable rules and engagement terms.
AI Efficiency Must Be Reflected in Reasonable Fees
Lawyers who bill clients hourly must bill for actual time spent working. AI-driven efficiencies should be reflected in fees. Lawyers may not charge clients for time spent learning general AI tools, and must clearly communicate how AI costs factor into billing.
Zuben's Approach
Zuben's credit-based consumption model provides per-run cost transparency. Every Constellation run logs its resource consumption, enabling firms to accurately allocate AI costs to specific matters and communicate those costs to clients.
Client Information Must Be Protected in AI Systems
Lawyers must know how AI tools use data and implement adequate safeguards. Formal Opinion 512 recommends obtaining informed consent before using client confidences in AI tools and warns that boilerplate consent in engagement letters will not be adequate. Use of shared AI tools may result in inadvertent disclosure of client information.
Zuben's Approach
Zuben provides complete logical and physical isolation of client data. No model training on client data. No data commingling between matters or firms. Ethical walls enforced by a Permission Graph with RBAC and ABAC controls. AES-256 encryption at rest, TLS 1.2 and above in transit. Bring-your-own-key available for enterprise deployments.
AI Outputs Must Be Verified Before Submission to Courts
Lawyers must carefully review AI outputs to ensure assertions made to a tribunal are not false or misleading. AI "hallucinations" — fabricated citations or mischaracterized holdings — must not form the basis of legal arguments. Lawyers should be aware of applicable local rules regarding disclosure of AI use.
Zuben's Approach
Zuben's architecture is specifically designed to reduce — though not eliminate — the risk of inaccurate citations and mischaracterized holdings. The pipeline includes two verification gates: a pre-draft structural validation and a post-draft proposition-faithfulness audit that checks whether each cited authority actually supports the proposition attributed to it. These safeguards significantly reduce hallucination risk, but they do not replace the lawyer's independent obligation to verify all citations and legal assertions before submission to any tribunal. Lawyers remain fully responsible for the accuracy and completeness of all filings.
Firms Must Establish Clear Policies for AI Use
Managerial and supervisory lawyers must create effective measures to ensure that all lawyers and nonlawyer assistants — including AI tools — conform to the rules of professional conduct. Clear policies regarding AI use must be established and enforced.
Zuben's Approach
Zuben provides tools to support supervisory compliance, including firm-level administrative controls, per-matter audit trails, and role-based access management. Every interaction is logged with query, user, role, matter, sources used, sources withheld, output classification, and model version. These capabilities are designed to facilitate — not replace — the supervisory structures that firms must independently establish and maintain under their ethical obligations.
State-Level Guidance
In addition to ABA Formal Opinion 512, numerous state bars and courts have issued their own guidance on AI use in legal practice. As of 2026, states including California, Florida, New Jersey, New York, Pennsylvania, Texas, and Illinois have released ethics opinions or guidelines. Many courts have also adopted local rules requiring disclosure of AI use in filings. Zuben's Forum Profile feature automatically identifies and applies forum-specific rules, including AI disclosure requirements, to every litigation work product.
Our Ongoing Commitment
As generative AI continues to evolve, ethical obligations will evolve with it. Zuben is committed to updating our architecture, policies, and practices in response to new guidance from the ABA, state bars, and courts. We participate actively in the legal AI ethics conversation and welcome dialogue with firms, bar associations, and regulators about responsible AI deployment in legal practice.
Questions about our approach to responsible AI? Contact us at support@zubenai.com.
Verification Is Not Optional
See how Zuben's architecture addresses your firm's ethical obligations.