Kilpatrick's Steve Borgman and Jordan Glassman recently presented on the various generative AI technologies available, with an emphasis on large language models such as ChatGPT (and others that are similar), at Kilpatrick’s annual In-House Counsel Summit in Houston. They also discussed the risks and liabilities that organizations may be exposed to as their personnel use generative AI technologies more and more. Additionally, Steve and Jordan addressed what guardrails might be used to help manage the risks that come with using generative AI.
Key takeaways from their presentation include:
1. Generative AI has arrived. ChatGPT, DeepSeek, and many other Artificial Intelligence (AI) services are now widely available. Whether you know it or not, at least some of your employees have been using AI services like ChatGPT and others for work.
2. Generative AI holds promise, but also involves risks. Some of the risks include concerns about privacy, confidentiality, security, biased outputs, infringement and ownership of intellectual property rights, outputs with hallucinations or manipulated results, and general fear, uncertainty, and doubt due to the fast pace of technological and legal developments in AI.
3. Consider vetting AI platforms for use for company business. What terms apply to your use of AI? Do those terms allow the use of the AI output in the way the company wants or needs to use the output? Does the AI platform provide results that can be trusted, or is the output full of hallucinations? Does the AI service treat the information provided in confidence or will that information be shared and possibly provided as output to other users?
4. Instead of an outright ban on AI, consider allowing the use of AI subject to guardrails to manage your risks. One way to help manage AI risks is to develop and adopt policies and procedures for the use of AI. For example, which AI services are allowed? Are any forbidden? What company information may be shared with the AI platforms, and what subject matters should not be shared? What documentation should be retained when an AI service is used?
5. Do your existing insurance policies adequately protect against the risks that come with AI? Such questions often involve a consideration of how the AI output may be used and the risks that such use may involve. For example, use of AI-generated artwork might involve a risk of copyright infringement, but use of AI-generated software to provide a medical diagnosis involve quite different risks and may implicate different insurance coverage.
