CBA Record January-February 2026

LPMT BITS & BYTES BY KEVIN THOMPSON A Quick Overview of the ARDC’s AI Guide D ecember’s Tech for the Holidays seminar fea tured a spirited discussion of generative AI-driven search engines such as Perplexity. However, we also started talking about the ARDC’s recently released Illinois Attorney’s Guide to Imple menting AI. Since implementing an AI search engine will require a solid understanding of confidentiality, security, and ethical issues for a legal practice, I decided instead to provide a top-level look at some of the key provisions of the Guide. Appendix 2 of the ARDC’s Guide provides a checklist of common questions to help evaluate both general generative AI tools (like Perplexity) and legal-specific ones. The Guide distills the key legal, ethical, and technical considerations into a practical review of the Terms of Use. It does not ask every possible question, however, and should only be considered as a starting point for reviewing any AI tool. For implementation purposes, third-party managed service tools (such as Perplexity) are not directly controlled by the lawyer or law firm. Platforms in which AI process ing occurs on infrastructure operated by the model provider, not the law firm, present a security tradeoff—a user doesn’t have direct control over their data, but they also do not have to be directly responsible for system security. In such a third-party system, it is recommended that the user take proper safeguards to control what data are shared with the system. By default, free versions of some AI search engines (e.g., Perplexity) use your data to train their models and improve their service. However, you can generally opt out of this practice by subscribing to a paid plan. Based on the questions raised in the ARDC Guide’s Appendix 2, the paid plans are more likely to be implemented in a legal practice. As discussed in the Guide, many of the AI implementation options will depend on the nature of your practice and what types of data must be kept confidential (e.g., if the data can be classified as Sensitive Personal Information—as opposed to merely Confiden tial Information—more safeguards are needed). This will inevitably impact the type and These tools give users the option of selecting differ ent large language models and can quickly streamline broad web searches into precise and verifiable results, although they are not legal-specific. I had intended to use this column to help explore whether an AI search engine might make sense for your practice.

scope of client consent needed to use such a platform. Individual clients may not be able to opt out of certain AI-powered tools if they are used across a practice. This is more likely for tools such as Perplexity, which reach across services (e.g., email) and integrate web browsing into those services. In such cases, the Guide recom mends focusing on enterprise-level safe guards that can be implemented across your firm’s infrastructure. Remember, the Model Rules do not require lawyers to be personally knowledgeable about every aspect of technology, but they are respon sible for being sufficiently competent to use it safely. It would be prudent to have discussions with an IT professional to see what is possible in your network environ ment and review industry standards. Appendix 3 of the Guide is a sample Notice of Artificial Intelligence Practices template. This can be adapted for use in engagement materials or even possibly disclosed on firm websites. When more sensitive data is involved, enhanced expla nations and informed written consent may be prudent. Appendix 4 is a sample AI Use Policy that a firm could modify and implement. It explains what tools are approved to be used within a firm and ensures that users agree to follow it. Without a policy in place, or without having an enforceable one, a “Shadow IT” infrastructure could operate within a firm without the knowl edge of anyone responsible for safeguard ing the IT infrastructure. Overall, the Guide’s practical tips can help your firm develop a sound AI Use Policy that is tailored to you and your cli ents. It is not a one-size-fits-all approach, but it makes the process straightforward and much easier to implement. The ARDC has provided a valuable resource for Illinois attorneys.

Kevin A. Thompson heads the intellectual property practice at Levin Ginsburg, where he is a partner; he also chairs the CBA’s Law Practice Management and Technology Committee and co-chairs the International and Foreign Law Committee. He receives no compensation by vendors for products mentioned in this column.

50 January/February 2026