AI Acceptable Use Policy

O’Melveny is committed to fostering an innovative environment and leveraging cutting-edge technology, including artificial intelligence (AI) and Generative AI (GenAI), to enhance client services and improve operational efficiency. However, the adoption of AI technologies introduces unique risks to client confidentiality, professional responsibility, and system integrity. This policy defines the authorized use and strict restrictions on AI tools within the firm.

1. Authorized vs. Unauthorized AI Tools

The firm separates artificial intelligence tools into two distinct categories: Approved Enterprise Tools and Prohibited Consumer Tools.

• Approved Enterprise Tools: Users may only utilize AI platforms and tools that have been officially vetted, security-cleared, and licensed by the O’Melveny IT and Security Compliance team. These include firm-hosted enterprise LLMs and licensed legal technology applications with data protection guarantees.
• Prohibited Consumer Tools: The use of free, public, or consumer-grade GenAI tools (e.g., standard ChatGPT, free Claude, Gemini, or public coding assistants) is strictly prohibited for any firm business or client-related work. These tools routinely retain prompts and output to train their models, presenting a severe risk of confidentiality breach.

2. Protection of Client Confidentiality & Privilege

The attorney-client privilege and duty of absolute confidentiality are foundational to O'Melveny's practice:

• No Sensitive Inputs: Under no circumstances may a user input client names, identifying case details, trade secrets, proprietary business information, or draft contracts into any external or unvetted AI tool.
• Data Boundary Compliance: For approved enterprise AI tools, ensure that all prompts and data submissions stay within the firm’s private tenant space as designated by IT.

3. Accuracy, Validation, and "Human in the Loop"

Generative AI tools are known to occasionally produce "hallucinations" (confident but incorrect statements of fact, citation, or law):

All AI-generated content—including legal research, briefs, correspondence, code, and administrative analyses—must be rigorously verified by a qualified professional before being used or shared. AI output must never be used as-is for legal filings or client deliverables. Every user is personally responsible and accountable for the accuracy and professional quality of any work they produce, regardless of whether AI tools assisted in its creation.

4. Intellectual Property and Licensing

AI models may generate material that contains copyrighted work or resembles proprietary code and intellectual property. Users must verify the copyright status and license compliance of any code, template, or copyrighted material suggested or generated by AI tools before incorporating it into work products.

5. Monitoring and Enforcement

O’Melveny reserves the right to monitor, log, and audit all prompts, inputs, queries, and outputs generated on firm-owned networks or when using approved firm AI platforms. Non-compliance with this policy, including the unauthorized input of client data into consumer-grade AI tools, will be subject to disciplinary action, up to and including termination of employment or contract.