The double-edged sword of lawyers using artificial intelligence

In-house lawyers especially should assess risks and rewards before implementing new legal tech

The double-edged sword of lawyers using artificial intelligence
Fernando Garcia

I have long advocated for the advancement and application of legal technology. I firmly believe that such technology will be instrumental in allowing lawyers to “do more with less” by increasing work efficiencies and removing the monotonous and laborious elements of legal practice, letting lawyers focus on the tasks that make their work more interesting and value-additive. Moreover, these advancements in legal technology may have important implications for Canada’s ongoing access to justice issues by potentially making legal services more affordable.

I have argued that not utilizing these tools may contravene the Law Society of Ontario’s Professional Code of Conduct. Rule 4.1-1 requires lawyers to provide services “in an efficient and convenient way.” However, it is becoming increasingly clear that while advocating for greater use and implementation of legal technology, lawyers ­– especially in-house lawyers – must exercise caution when using such technologies in the provisions of services for clients and their internal business partners.

For starters, programs such as ChatGPT function by drawing from a pool of past practices, precedents and other historical information. When the program was first introduced, people began using it for fun, such as to write songs, cards and stories. Then lawyers began to use this tool to conduct legal research and draft documents, from statements of claim to research memos to organizational policies.

While ChatGPT helps establish a baseline for legal arguments or documentation, the true measure of a successful lawyer is developing novel arguments, reading the room when dealing with members of the judiciary and juries, and effectively analyzing their clients' risk appetite. Excessive reliance on a tool that only looks backward will not provide efficient or effective services.

There is also the risk that, by relying on historical information, legal technology, like human beings, has the potential to be influenced by bias, ideologies, and moral judgments. Without being aware of and accounting for these elements, members of the bar who mindlessly rely on the output of these tools risk hurting clients, the public, and society. The reason for this is quite simple: legal technology is not immune from the concept that “garbage in” results in “garbage out.”

For in-house counsel, using these new tools also requires the application of precautions. For example, an in-house counsel may try to minimize costs and be more efficient by using a tool such as ChatGPT to draft sensitive correspondence, such as a statement of claim. The individual then proceeds to ask the tool to prepare a statement of claim whereby company X intends to sue company Y, in the courts of Ontario, for some specific contractual breach. Unfortunately, the fact that X intends to sue Y in Ontario for the breach of agreement is now part of the public domain. In using ChatGPT, has in-house counsel unknowingly disclosed a material matter publicly?

Another example would be if a lawyer or in-house counsel uses ChatGPT to draft or research a legal opinion on the likelihood of success of having X sue Y by providing details of the claims and facts, has counsel lost client-solicitor privilege over the legal advice produced as it too is in the public domain?

While there are ways to minimize the risk by not utilizing the name of the parties or information, these strategies and the management of the risks need to be identified and addressed in a potential workplace AI policy to would-be users. Training may be required to minimize the risks. Such a policy is essential.

Another demonstrative example is the use of online translation tools. In-house counsel and business team members may receive a contract, proposal, or letter from a regulator in another language, or they may use one of these online tools to translate commercially sensitive business plans, customer pricing, and confidential proposals. All these documents have now, without proper safeguards, become part of the public domain. This may breach confidentiality clauses in the agreements and raise troubling competition implications from the public release of commercially sensitive information. It is essential to clarify with the business and legal teams when these online tools can be used, how certain information must be redacted – or, better yet, when online tools cannot be used to translate business-sensitive documents.  

Yes, as competent lawyers, we must strive to keep updated on the development of legal technology and the tools available for providing more effective and efficient services. However, we must all be mindful of the consequences – and mitigate the risks these tools can introduce into our legal practice or within our organizations. These extraordinary advancements and programs are essential tools to help us become even better lawyers. Still, they can often prove a double-edged sword if not used appropriately and with adequate safeguards. Govern yourselves accordingly!