Not So Fast: ChatGPT May Be Great for Attorney Efficiency, but Don’t Leave Ethics at the Door
TikTok, who? ChatGPT – an artificial intelligence (AI) chatbot developed by OpenAI – is the fastest-growing app of all time, according to analysis by Swiss bank UBS. A natural language processing tool that allows human-like conversations on an infinite number of subjects, ChatGPT is trained to follow a prompt and provide a quick and detailed response, and numerous generative AI software programs are emerging alongside it. But with this capability comes risks, limitations, and ethical considerations significant to the legal industry.
The Model Rules of Ethics require attorneys to maintain “technological competency,” meaning they must keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology. Therefore, attorneys cannot bury their heads in the sand when it comes to ChatGPT because, in reality, their colleagues and clients are already using it and other programs like it, so it is essential that every attorney understand the risks and benefits.
The emergence of ChatGPT and similar generative AI programs present opportunities for automation and efficiency, but the technology is still in development and cannot be relied upon blindly without an attorney exercising his or her own judgment. ChatGPT gathers information obtained from the Internet, but it does not always present the most relevant data in generating responses and has been a frequent source of misinformation in its responses to queries. For instance, early attempts to query ChatGPT for case law often yielded cases invented by ChatGPT, which perhaps understood such queries as creative writing prompts rather than as legal research requests. The software now avoids providing case law altogether, which seems prudent given that ChatGPT was not designed to specifically evaluate questions of law or to mine and evaluate data specific thereto.
While ChatGPT can still prepare responses on a nearly infinite number of subjects, a lawyer’s ethical duty of competence should always take precedence over convenience. Attorneys must do their due diligence by verifying any answers that come from ChatGPT, comparing them to their own knowledge or conducting their own research from reputable sources. Importantly, ChatGPT is also limited in its ability to provide sources or evidence to support where it obtained its answers, so it is imperative to independently verify any answers generated by ChatGPT.
Another concern among users is that data cannot be retrieved or deleted once entered into ChatGPT, as it is stored on the servers belonging to OpenAI. Each time users chat with the AI, the conversations and user inputs are saved as an ongoing conversation thread to help improve the AI’s accuracy over time. However, the process in which personal information is used for data training may make ChatGPT non-compliant with certain privacy regulations, such as the European Union’s General Data Protection Regulation (GDPR), as one of the core tenets of the law is governing how companies collect and use data. Italy, in fact, had temporarily banned the use of ChatGPT in March 2023, citing concerns for privacy violations, the potential spread of misinformation, and the lack of an age verification system preventing minors from exposure to illicit materials. At the time of writing, OpenAI says it has fulfilled conditions Italy wanted satisfied by an April 30 deadline to have the ban on ChatGPT lifted, including adding information on its website about how it collects and uses data and adding a tool to verify users’ ages when signing up. OpenAI also offers the option to opt out if you do not want your data used for training, but this is a manual process that involves filling out a form.
Another area of uncertainty is the risk of violating someone’s intellectual property in using responses prepared by ChatGPT. According to ChatGPT, the user owns the copyright in the user’s prompts and the generated responses the user receives. However, it is unknown how and from where ChatGPT is obtaining its data. As noted above, ChatGPT scans the web to gather information from open data sources on the Internet. There is no way to verify that what is being generated is not going to infringe on someone else’s copyright; therefore, firms should advise their attorneys and clients to proceed with caution using any content developed by ChatGPT and should generally use such content for internal informational or educational purposes only.
By understanding the limitations of the AI technology and being mindful of the ethical considerations summarized above, there is potential that the ChatGPT tool can be used in a way that is both innovative and ethical. At present, however, attorneys must continue to critically assess and implement this technology responsibly to ensure they uphold their ethical obligations and protect client interests.