AI ETHICS Enterprise-level AI tools are deployed through organizational accounts with negotiated terms of service, dedicated security infrastructure, and administrative oversight. These systems typically operate in closed environments where data remains within the organization’s control. Examples include Microsoft 365 Copilot deployed through a firm’s existing Microsoft license, legal-specific platforms like Harvey or CoCounsel with enterprise agreements, and custom AI implementations hosted on private servers or secure cloud instances. Enterprise tools generally offer stronger confidentiality protections, audit capabilities, and compliance features 67 — but they require significant financial investment and technical infrastructure. Personal AI tools are consumer-facing products accessible to individuals through free or low-cost subscriptions. The free versions of ChatGPT, Claude, and Gemini fall into this category. While these tools provide powerful capabilities, they typically operate as open systems where user inputs may be retained, analyzed, or incorporated into future training. Terms of service for personal accounts rarely include the confidentiality guarantees that legal work requires. The dangers of personal tool use for client matters cannot be overstated. As discussed in Part II, entering confidential client information into open systems risks breaching Rule 1.6 obligations. Even seemingly innocuous queries may reveal case strategies, client identities, or privileged communications. A defender who uses a personal AI tool to quickly review their state’s rules on expert testimony during a break in trial, for example, may inadvertently disclose client facts to OpenAI and its contractors, potentially waiving privilege and violating confidentiality duties. Defenders should also understand that different tools excel at different tasks. Large language models like ChatGPT and Claude perform well for text generation, summarization, and analysis. Image generation tools like DALL-E and Midjourney can create visual aids for trial presentations. Specialized legal AI platforms may offer enhanced capabilities for document review, citation checking, and case law research. Matching the right tool to the task improves both efficiency and output quality. For defenders lacking access to enterprise tools, some organizations offer subsidized access, shared subscriptions, or resource pools for certain legal-specific AI tools. 68 And in some circumstances, careful use of consumer tools for non-confidential tasks — such as researching general legal principles or drafting non-case-specific training materials — may be appropriate, provided defenders rigorously avoid entering any information relating to client matters.
2.
Vendor Selection
Choosing appropriate AI vendors requires evaluation across multiple dimensions. Defenders must balance functionality against security, cost against capability, and convenience against compliance.
34
Parity in Practice: The Defender’s Duty to Ethically Use AI
Made with FlippingBook Online document maker