Originally interviewed in 2024 for the Marcus Evans Chief Litigation Officer Summit.
“CLOs need to fully vet the risks associated with using generative AI (“AI”) tools in the business, including the legal practice. They should be involved in setting internal policies for AI usage, to take the advantages of AI benefits and avoid the pitfalls,” said Nancy Wolff, a Partner at Cowan, DeBaets, Abrahams & Sheppard LLP.
“Employees need to know how to use AI safely, what issues can potentially arise, and why human involvement is a must when generating new content in order to maintain as many rights as possible,” added Scott Sholder, also a Partner at the firm..
Cowan, DeBaets, Abrahams & Sheppard LLP was a sponsor of the Marcus Evans Chief Litigation Officer Summit 2024. In this interview, Wolff and Sholder discussed the risks and opportunities surrounding generative AI, including responsible AI usage, internal policies, copyright protection and potential litigation risks.
Wolff: Generative AI has so much potential. It is great at making tedious tasks easier. However, in the rush to get AI products out in the market, many vendors have not focused enough on some very important issues. Their thinking has been to launch first, and work out legal issues later. Companies need a good understanding of how employees are using AI and on what platforms to create appropriate corporate policies.
Sholder: There are many potential potholes, but many organizations have started using AI without fully assessing the risks. This is not the best practice; diligence, research, and obtaining an understanding of how the technology works is essential.
Wolff: AI tools are constantly improving, but when they are embedded or incorporated in the business, CLOs need to instruct their teams to be careful not to feed any confidential client information on a publicly available platform that will use the information as part of its learning. Employees should not rely on the results of any AI generated content without properly reviewing it. Human oversight is a must. The data might not be accurate, especially in fields where the source material for training is limited. When AI doesn’t have an answer, it simply creates its best guess.
Wolff: The risks are similar. When they use AI for discovery and document review, the output will not necessarily be accurate. It is best to use AI tools that are specifically designed for the litigation market, that have built-in security features, and not taken from large language models that are generally scraped from the internet. Such curated AI platforms will become very valuable for keeping litigation costs down.
Sholder: Putting together internal policies that everyone in the organization must read and potentially sign can set the stage for responsible use and risk mitigation. The policies should cover how everyone can use AI responsibly and ethically. CLOs need to consider how the company can avoid getting sued and having to sue based on AI usage, and internal policies can help achieve these goals.
Wolff: The policies may need to be unique to different business units or departments and how each may use generative AI. These policies should be continually reviewed and updated, as the technology and regulations change. For example, certain industries may be more regulated based on privacy issues, such as the banking, medical and health industries. Everyone understanding how the technology works will go a long way in avoiding risks. They need to protect trade secrets and make sure they are not inadvertently being added to any training models that competitors might utilize.
Wolff: It depends on each company, but for example, a publishing company may see opportunities to create illustrations or some text with AI, but the raw output will not be protected by copyright. It could also be infringing depending on what was used in the model. Those using AI should save the prompts used and the iterations of a work to show the human involvement in the final product. In general, it is best to use AI for outlining and ideation, and to have a human touch added to the underlying baseline AI output. This is important if the company wants ownership of the output for itself or for its clients.
Sholder: With AI, there is a lot of potential for litigation risks, including copyright infringement and contract or license breaches. There are many angles CLOs must cover before litigation even becomes likely. When litigation does occur, it is critical to preserve records and documents relating to the case, including any training material, prompts, outputs, and communications.
"Companies need a good understanding of how employees are using AI and on what platforms to create appropriate corporate policies"
View Nancy Wolff on Linkedin here.
View Scott Sholder on Linkedin here.
Find out more about the upcoming Chief Litigation Officer Summit.
For more information, please contact: Kyriakos Xenophontos, kyriakosx@marcusevanscy.com