AI ETHICS C. Additional Policy Recommendations
Beyond office-level policies, defenders should pursue collective approaches to AI governance. Moreover, no model policy can anticipate every organizational context, and this one is no exception. A large law firm deploying AI across dozens of practice groups faces a fundamentally different set of risks than a solo practitioner using an AI drafting tool for correspondence, and a policy calibrated for one will almost certainly be over – or under-inclusive for the other. The provisions will require careful tailoring: what counts as an acceptable use case, how broadly or narrowly to define permitted tasks, and where to require human review are questions that turn on facts specific to each organization and its size, its practice areas, its client base, and its existing data governance infrastructure. Organizations should treat the framework offered here as a starting point for internal deliberation rather than a finished compliance solution and should expect that meaningful implementation will require revisiting and refining these provisions as their own AI use matures.
1.
Timeline for Adoption
Although this white paper recommends defense lawyers and offices should implement formal AI use policies within one year, it recognizes that not all defender offices, firms, or practices operate the same. Indeed, it may be impossible to implement an AI use policy for some systems within a year. At the very least, setting a deadline for development and implementation of an AI policy is essential. Given the pace of AI adoption and the risks of unguided use, delay exposes clients and practitioners to unnecessary harm. Interim measures should begin immediately. Even without a comprehensive written policy, defenders can establish informal guidelines, restrict use of non-secure and/or personal AI tools, and ensure that any AI use receives appropriate oversight. The model policy provided in Appendix A offers a template that defenders can adapt to their specific firm or office.
2.
National Coordination
NACDL is well-positioned to coordinate AI policy development across the defense community. Consistent policy language may help reduce confusion when defenders collaborate across organizations. Shared resources — including model policies, training materials, and vendor evaluations — reduce the burden on firms and offices, especially those without adequate resources to develop policies and procure tools from scratch. NACDL’s Task Force on Artificial Intelligence has compiled a list of helpful Champion® articles and other resources about adopting AI in criminal defense, including commentary from the NACDL President, techniques for mastering prompt engineering, and third-party resources (e.g., books, articles, webinars) that may be helpful to defenders. 78
45
Parity in Practice: The Defender’s Duty to Ethically Use AI
Made with FlippingBook Online document maker