Issue 60 

Lawyers are hearing: AI will be a gamechanger for your law practice. 

At the very same time, lawyers are hearing: lawyers are getting sanctioned for using AI. 

This tension between opportunity and risk has left many lawyers confused about how to use AI safely, or whether they should use it at all.  

There are hundreds of AI tools available to lawyers, which adds to the overwhelm and confusion. Arguably, the AI tool with the most name recognition and visibility is ChatGPT. Many lawyers have picked ChatGPT or another general-purpose AI tool like Claude or Gemini as their first AI tool. It’s easy to understand why: general-purpose AI tools like ChatGPT, Claude, and Gemini are well-known, can perform a wide range of tasks, and can be accessed relatively inexpensively.  

Yet, there is conflicting messaging about whether it is safe for lawyers to use general-purpose AI tools like ChatGPT for legal work. Lawyers are left wondering whether you can actually use ChatGPT (or any other general-purpose AI tool) for legal work, or if it’s too risky.  

The Short Answer: Yes in Some Cases, But Consider the Big Picture Before Deciding 

Yes, lawyers can safely use general-purpose AI tools like ChatGPT in a law practice in some contexts. However, that “yes” is qualified, because not all possible uses of general-purpose AI tools like ChatGPT are well aligned with lawyers’ professional responsibilities.   

How You Use ChatGPT Matters

In Formal Opinion 512, the American Bar Association (“ABA”) cautioned lawyers to confirm that an AI tool will preserve the security and confidentiality of the information disclosed if the AI tool will receive data related to a lawyer’s representation of a client. AI tools can be used within a law firm for a very wide variety of tasks, ranging from tasks like marketing that are unlikely to involve client information, to tasks like brief writing, which are much more likely to involve client information. Given the wide range of possible uses, lawyers need to consider all the ways an AI tool like ChatGPT could be used within their law firm (including accidental or unintended transmission of client information to the AI tool) in conjunction with the confidentiality and security safeguards offered by the AI tool.   

Some law firms have decided to allow employees to use general-purpose AI tools like the consumer-grade version of ChatGPT, but prohibit employees from sharing confidential client information with the AI tool. Unfortunately, this type of policy underestimates the real risk of human error. Sharing confidential information with an AI tool that will not safeguard the information is an error that cannot be undone. A policy that relies on human beings to be perfect 100% of the time is not likely to be considered reasonable in the aftermath of an inadvertent disclosure of confidential information. Lawyers need to consider not only how they intend for an AI tool to be used at their law firm, but more importantly, how that AI tool might be used in ways they do not intend or desire.  

The Version of ChatGPT You Use Matters

When it comes to the confidentiality and security safeguards of an AI tool, it’s important to know that general-purpose AI tools like ChatGPT commonly offer different tiers of service, ranging from free and low-cost consumer plans, to more expensive business and enterprise plans. In general, as the cost of a general-purpose AI tool rises, the confidentiality and security safeguards often also increase. Lawyers should investigate the terms of use and any other security or privacy policies of the specific version of the AI tool they are considering using in their law practice to determine if the confidentiality and security safeguards offered align with their professional responsibilities.   

Why Lawyers are Getting Mixed Messages About Using AI Tools Like ChatGPT 

There is an overwhelming amount of information for lawyers to wade through as they decide which AI tool(s) to use in practice. Lawyers have hundreds of choices of AI tools, which offer different features and levels of privacy and security protection. Because the market for AI tools is so new, there is a lack of clear consensus surrounding which AI tools are actually worthwhile for lawyers. There are a lot of lofty marketing claims and hype, and significantly less independently sourced information about how AI tools for lawyers operate in reality.   

Meanwhile, governments and bar associations are grappling with how to best regulate AI. AI advancements are occurring rapidly, which adds layers of complexity for regulators who must contend with a moving target and craft regulations that will not be outdated before they are enacted. Additionally, in some instances, there is debate over whether new laws and rules are necessary, or whether existing laws and rules can be applied to AI.  

While AI rules and guidance are evolving, there are weekly headlines about lawyers getting sanctioned for filing documents containing hallucinated cases. Additionally, lawyer discipline for AI-related misconduct is not limited to court sanctions, as bar association investigations and discipline are becoming more widespread.   

Risks of Using ChatGPT in Legal Practice 

Lawyers who want to reduce their risk of violating their professional responsibilities while using an AI tool like ChatGPT in legal practice should familiarize themselves with the following risks:    

Confidentiality Risks Associated with AI Use

Not all AI tools offer the same data confidentiality and privacy protections. Oftentimes, the consumer-grade versions of general-purpose AI tools like ChatGPT (which can be free or relatively low-cost) do not offer robust data security. Some AI tools may default to using the data you input to train their AI models, or may use your data for other commercial purposes. Inadvertently entering confidential client information into an AI tool that will not safeguard the data is like ringing a bell: it cannot be unrung. Once confidential data is entered in an insecure AI tool, it cannot be undone. In contrast, some business or enterprise-grade general-purpose AI tools, as well as many legal-specific AI tools, offer greater data confidentiality and privacy protections that may better align with a lawyer’s professional responsibilities. Lawyers have a responsibility to evaluate the terms of use and other security/privacy policies associated with an AI tool before using it for legal work.  

Additionally, leaders of law firms should consider the risk that employees may be putting confidential client or firm data at risk by using AI tools that have not been vetted by the firm. This issue is commonly known as the use of “shadow AI”. Firm leadership should provide routine training to employees to emphasize the importance of safeguarding confidential client and firm information in relation to AI use, and should enact a clear policy regulating workplace AI use. 

AI Hallucinations and Accuracy 

Lawyers should be aware that AI tools, including general-purpose AI tools and legal-specific AI tools, all are at risk of producing inaccurate information, sometimes referred to as hallucinations. Some real-world examples of AI hallucinations impacting lawyers include: 

• Fabricated case citations;  
• Misstated case quotations; and  
• Misleading summaries of cases.  

Overreliance on an AI Tool 

Given the risk of hallucinations in output from any AI tool, lawyers must be skeptical of the accuracy of all AI output, and always verify accuracy before relying on the output. Overreliance on the output of an AI tool can lead to filing court documents including inaccurate information, which can lead to sanctions and other disciplinary action. Lawyers who ultimately sign documents drafted by others must also be aware of the possibility of AI use by their collaborators, and should remember their independent duty to verify the accuracy of a filing before signing.  

When ChatGPT Can Be Useful for Lawyers 

A consumer-grade general-purpose AI tool like ChatGPT could prove useful to lawyers in certain contexts that do not involve confidential information. A few examples include: 

• Brainstorming responses to a filing from opposing counsel; 
• Summarizing data that is not confidential; and 
• Firm marketing and SEO efforts.  

When Lawyers Should Be More Cautious About Using ChatGPT 

Law firm leadership should consider the potential unintended consequences of permitting firm-wide use of a consumer-grade general-purpose AI tool that does not offer adequate security and privacy protections. This amounts to relying on every person in the firm to always remember to avoid putting confidential information into the AI tool when performing their work. It’s simply less risky to instead use an AI tool that offers confidentiality and privacy protections that align with lawyers’ professional responsibilities. 

Additionally, lawyers should always be cautious of the risk of hallucinations in AI tool output, no matter which AI tool they use.  

How ChatGPT Compares to Legal-Specific AI Tools 

ChatGPT and other general-purpose AI tools are attractive to lawyers because of their relatively low price point, and the wide variety of tasks they can perform. Additionally, general-purpose AI tools are currently competitive with legal-specific AI tools when it comes to output accuracy.    

In contrast, legal-specific AI tools are typically more expensive than free and low-cost consumer-grade general-purpose AI tools, but also often offer greater data protections, as well as other potential benefits that appeal to lawyers. For example, some legal-specific AI tools have automated legal prompting, or do not require prompting at all, while general-purpose AI tools can require an investment of time to figure out how to improve output through legal prompt engineering. Legal-specific AI tools also often include workflow integration benefits that lawyers find valuable.  

A Better Approach: Evaluate AI Tools Before Using Them 

Instead of jumping straight into experimenting with a well-known AI tool in your law practice, take time to understand precisely what the tool offers. This is where many lawyers go wrong: by failing to evaluate an AI tool to understand its benefits and risks before starting to use it. Moreover, lawyers can benefit from following a structured process when choosing AI tools, which includes considering the technology improvement priorities of your law firm before picking specific AI tools to evaluate further.   

What You Should Do Next 

Law firm leaders should begin by zooming out from considering specific AI tools, and should first ensure: 

• The firm has a policy regulating workplace AI use; 
• All members of the firm receive routine education about the risks associated with AI use; and 
• The firm’s technology improvement priorities have been identified.  

If you want a step-by-step framework for evaluating AI tools, including how to identify technology improvement priorities, assess risk, compare options, and implement tools responsibly, you can learn more in my 2-hour CLE: How to Pick the Best AI Tools for Your Law Practice.  

Or if you’d prefer a more comprehensive resource that also includes a curated directory of over 250 AI tools for lawyers, check out A Lawyer’s Practical Guide to AI.  

Thanks for being here. 

Jennifer Ballard
Good Journey Consulting