Generative AI is changing how personal injury firms review documents, draft work product, and manage cases. But the professional rules haven't changed — shift from “using” AI to “governing” it.
Competence now includes algorithmic competence. Understanding a tool's underlying logic, limitations, and biases is part of the duty of competence (Rule 1.1). That means vetting vendors for security certifications like SOC 2 and HIPAA compliance, and confirming a zero data retention policy before feeding in client information.
Confidentiality doesn't pause for convenience. Rule 1.6 still applies. Staff need training on how AI tools handle data so one client's confidential information never bleeds into another client's file — and firms should be ready to explain AI use to clients who ask (Rule 1.4).
Candor is non-negotiable. Beyond the well-publicized risk of AI "hallucinating" fake cases, there's a subtler danger: sycophancy — AI's tendency to tell you what you want to hear. An advocate has to prompt the tool to find weaknesses, not just build the argument they already believe.
Supervision must be documented, not assumed. Rules 5.1 and 5.3 require supervising staff who use AI. A verification log — who checked the sources, what was manually confirmed, how the prompt was structured — turns vague oversight into an auditable process.
The algorithm is a tool. The advocate remains the steward of justice. Firms that pair AI's speed with rigorous, SOP-driven supervision get the benefit of the machine without losing the soul of the practice.