Mintz On Air: Practical Policies — Real vs. Robot: Bots in the Boardroom

What happens when AI has a seat in the boardroom?

In the latest episode of Mintz On Air: Practical Policies, Real vs. Robot: Bots in the Boardroom, host Jen Rubin is joined by Member Melanie Levy for an unscripted conversation on AI’s evolving role in corporate governance.

They explore many key questions boards face today with AI’s rapid escalation and widespread adoption — how to discharge fiduciary duties that apply to humans who use AI tools, the risks to confidentiality, technology adoption without appropriate policies, and the potential for diminishing the protection of the business judgment rule. They also discuss whether banning AI in the boardroom or requiring fluency tests for board members is realistic, and why AI can enhance decision-making but not replace human judgment.

A practical listen for board members, in-house counsel, and governance professionals navigating AI with accountability and oversight.

Mintz On Air: Practical Policies — Real vs. Robot: Bots in the Boardroom Transcript

Jen Rubin (JR): Welcome to the Mintz On Air: Practical Policies podcast. Today’s topic: Real vs. Robot: Bots in the Boardroom. I’m Jen Rubin, a Member of the Mintz Employment Group with the San Diego–based Bicoastal Employment Practice. I represent management executives and corporate boards. Thank you for joining Mintz On Air. If you have not tuned in to our previous episodes and would like to access our content, you can visit us at the Insights Center at Mintz.com or you can find us on Spotify.

Today I am pleased to host my partner Melanie Levy. We are going to be chatting about AI in the boardroom. I’m not just talking about using AI agents in the boardroom for boardroom work, although we will be talking about that. I’m keenly interested in discussing whether AI can not only enhance board performance, but whether AI actually can replace human board members.

Melanie, before we dive in, can you tell our listeners a bit about your practice and your experience with the various rules and regulations applicable to decision-making in the boardroom?

Melanie Levy (ML): Thank you for having me, Jen. I am a corporate attorney at Mintz, and one of my main roles is as a capital raising strategy attorney. I am also typically the attorney in the boardroom for most of my clients. I sit in their board meetings. I advise them on corporate governance, and on strategic and important decisions affecting the corporation and their shareholders.

JR: Well, your background is ideal for what we’re going to dive into today.

Fiduciary Duties in the AI Era

JR: The first topic I’d like to discuss are the rules of the road around using technology in the boardroom. Most of us are familiar with the concept of a board member’s duty of care, and the duty of loyalty to the organization. But Melanie, can you lay out for us generally what these duties require and why they might be applicable to the use of technology in the boardroom?

ML: At a high level, there are two basic duties — I’m going to focus on Delaware because that’s where most companies are incorporated — that board members for Delaware corporations owe to the company and to their shareholders. One is the duty of care, and it is a duty that requires directors to make informed decisions and to act with the care that a reasonably prudent person would exercise under similar circumstances. That’s a lot to unpack, and there’s a lot of flexibility there. There’s great deference given to directors in exercising that duty of care. They are allowed to rely on their business judgment. Courts will defer to them. Courts are not in the business of second-guessing directors or officers on the decisions they might make, in their good judgment, that would be in the best interest of shareholders.

The other duty is the duty of loyalty. It requires that you act in good faith and in the best interest of the corporation, not in your own interest or the interest of a third party.

How does technology, and AI in particular, implicate these duties? This is an important concept that I think is going to flow through a lot of what we talk about today. But Delaware’s fiduciary framework, the case law, the stock exchange, the SEC — every single part of these duties relies on the fact that it is a human being performing these duties. Delaware expects that it is a human being exercising those duties. If you want to use AI in your boardroom, is it helping you or is it replacing you? That’s a key aspect of the duty of care.

From a duty of loyalty perspective, everything is discoverable. AI is in writing. You’ve probably heard your corporate attorney tell you, “We don’t take notes in board meetings. We destroy all of our notes.” Imagine leaving a thought process and written record of your decision-making and potentially of the company’s secrets in a third-party platform.

JR: A platform that may not be secret and may not be protected.

Bridging the Knowledge Gap on AI

JR: Modern governance requires some level of adeptness with technology. It’s probably fair to say that a board member at a fundamental level needs to understand how to operate email, for example. And some board members may have different levels of comfort when it comes to harnessing technology to help fulfill the board member’s responsibilities. I’m going to throw out some scenarios for you on the subject so we can better understand how those responsibilities can be discharged, and whether technology can help or hurt.

Let’s start with this. Let’s assume, Melanie, that you’re at a board meeting. The meeting starts. Ms. Board Member asks the chair to include as an agenda item a rule that prohibits board members from using AI for any reason in connection with the discharge of that person’s board duties. And let’s assume this particular motion and approach comes from a defensible place. Let’s say that this board member is worried that confidential information might be disclosed through AI. Can you speak to that first?

ML: If something like that were to come up in your boardroom where somebody is passing a strong ultimatum, the first reaction should always be to ask: why? What is their concern? That’s always question number one. A good rule of thumb is that when somebody has a strong feeling, ask them a question, because it’s better than assuming.

If I were a board member or I were advising the board, I probably would ask why the person is concerned about using AI. The answer might be that they’re worried about privacy and confidentiality. Perhaps they think what they’re getting out of the bot or the software or the service is inaccurate, and they want to ensure that human interaction and human judgment, which the law requires, is being exercised.

What I would suspect would come out of that conversation would be that there is value in AI. In fact, I suspect in the not too distant future, it might even be considered not to be a reasonable standard to avoid the use of AI entirely. In the near future it may be considered that it isn’t exercising reasonable care to avoid the use of AI in all situations, and to have a complete bar on AI. But where it does make sense is when you’re asking the question, “You want us to not use AI, but what if we have AI that we have contracted with that’s built for us, that we have a provider we have agreements with, and licenses, so that confidentiality is maybe somewhat addressed?” And then what is it that you don’t want it doing and what do you want it doing? I think that’s a reasonable discussion to have as a board, because the fact is it’s not only what is the board doing, but your employees are asking themselves the same question. So you don’t want to leave that unsaid. But I think an absolute ban is not realistic. It’s not realistic to be silent on the subject or to say, “I never want them to use AI.”

JR: Let me segue into this because I think this is a fair question. I’m not going to paint a broad brush and say that people of a certain generation are afraid to use technology, but let’s assume for the moment that the board member who raised that agenda item actually doesn’t understand AI, doesn’t understand how it works, isn’t sure they’re going to be able to get up to speed on this technology train. I think this notion of asking questions is helpful. But let me go a step further, because I think it’s a fair assumption with many individuals. I think many of us are feeling that way. We don’t really understand AI. We don’t understand how to use it. What if the board member wants to propose some bylaw amendments that require all board candidates to take a test requiring proof of knowledge of AI fundamentals as a condition to be considered for a board seat?

ML: I would question that. I would question whether the most important attribute of a board member, their most important value that they bring to the table, is their ability to utilize AI. Would you really be in a situation where you had a preeminent board member, a person who was renowned for corporate governance — let’s say Jack Welch. Mr. Welch is going to come into your board room and you would disqualify him because he may not know about AI? It’s also a myopic question in that it leaves out the part that board members bring different experiences to the table. So if I were a drug development company, I would not want to lose the person that had FDA expertise because they may not also be an expert with AI. Part of working on a board is working as a team. Not every person on a board has to have that basic level. And if the board were so concerned, maybe a more reasonable thing to do would be to have an AI expert come in and talk to them about it, to get educated on it. Maybe there’s continuing education on AI. But setting as mandatory criteria for being a board member that you have to have a basic understanding of AI — I would be skeptical of that at this point. That said, it is fast approaching where it is going to be considered that you do not exercise or cannot exercise your fiduciary duties appropriately without using available technology and using modern resources. But that’s different from saying, “I’m going to ban you because you don’t know about this.”

JR: In-house counsel, corporate counsel, particularly those who are responsible for governance-related matters, may want to think about the type of training and policies that need to be implemented for the board member, in order to give this support. So instead of this myopic view that we’re going to ban it, let us provide the support so that we can have the board members work together collaboratively. Maybe a board member who’s an expert in technology, really understands AI, can help support the other board members. It seems to me that would be a priority for the board agenda to get that training.

AI as Advisor, Not Decision-Maker

JR: Let’s move on to another scenario. Let’s assume a board member in preparation for a regular meeting has run different scenarios for an important agenda item. Let’s call this agenda item whether or not the corporation should enter into a particular market. And let’s assume that whether the company should enter into the particular market is going to have material business consequences. And let’s assume that the board member has asked some of these questions of an AI agent. The board member comes to the meeting and advocates for the AI agent’s solution, without identifying that the board member has used AI to come up with that solution. What do you think of that?

ML: Here’s the issue. It is something I worry about, quite candidly, because the human impulse might be to just use ChatGPT to answer every question. We are so used to googling things — nobody ever sits with an unanswered question anymore, you just put it in Google. The problem though, in putting sensitive business information into something like ChatGPT, is that usually the decision to go into a new market to expand is a strategic decision. It’s a decision that if you’re a public company, it is probably material nonpublic information. It is possibly a trade secret of the organization. It’s about strategy. I think one thing to be concerned about is that you would not want to put all of your strategy into something like ChatGPT, just from a security and confidentiality perspective.

The other thing I would bear in mind — and Delaware law doesn’t speak to this directly, because Delaware law doesn’t really talk much about AI because it didn’t exist when a lot of these statutes were written — but Delaware law allows you to rely on a report. It allows you to rely on information from your executives, from individuals. You could not simply say, “I’m going to allow ChatGPT to make the decision for me.” It sets a bad precedent, too, because everything that board member puts into ChatGPT is discoverable. So if somebody said, “This was a major strategic decision, we need to see that there was appropriate due care,” and the only thing being shown in that ChatGPT discussion was that the person asked the question and was like, “I guess I’ll do what ChatGPT says,” that could be a problem. Again, AI can help you. It can be a wonderful assistive tool to gather information, but it is not something that should ever be accepted blindly.

Continuing this scenario, if the person being asked to show appropriate due care doesn’t say anything, well, sometimes you have to force the conversation about things, and this is probably a situation where we need to force the conversation. If you don’t have an AI policy at your company, if you haven’t decided how you want your employees to use AI or how you would like your board members to interact with AI, you’ve still essentially decided. You’ve decided that you have no policy, which means employees are probably using it in ways that you would not like, that are not good for the company, that are not good for the shareholders.

One key point where Delaware law has been clear in the past is the failure to exercise judgment. The failure to consider this use of AI — that is a breach of your fiduciary duty. You don’t get deference for a business judgment that you didn’t judge. If you don’t consider it, then the question will be, why didn’t you consider it? Did anybody ever think to ask or talk at the board level about how we should use AI? What do we believe our reasonable policy should be?

JR: I want to go back to what you said earlier about asking questions. It seems to me that that is the most appropriate method to tease out whether or not somebody has a basis to either support, provide details for, give the underlying analysis for a proposed course of action. And if somebody is unable to do that, that might itself disclose that the person didn’t actually go through that hard process of thinking things through. If there’s a suspicion by other board members that somebody’s information or work is not fully thought out, I think asking those questions can help discharge those duties. People need to be prepared to back up some of these suggestions. I think that’s a practical takeaway from this.

Can a Bot Be a Board Member?

JR: Let me ask you this question. The board meeting commences and a board member asks the chair to recognize the AI bot that the board member has brought to the meeting.

ML: It’s clear that merely making the AI bot a member of the meeting is— No. No. You could ask it questions, but the AI bot being given the floor on its own without direction of a board member, that is not Delaware law. Delaware law expects human beings to have the floor, not an AI bot. Again, it comes back to, is the AI assisting you in exercising judgment, or are you just turning it over to AI entirely? I have to say, I don’t think most board members would do that. They might allow AI to be in the meeting — which again has its own problems from a discoverability perspective. They may allow AI to be used to provide reports and resources. But they would never defer an important business decision wholesale to AI or give it the floor in the meeting. The bot is not a human.

JR: Presumably when those materials and resources are shared with other board members, the derivation of those materials is made clear. For example, we asked management to prepare reports, or our auditors did. In the same way that when we use AI in our law practices, we say to AI, show us your work. We’re not just going to accept what it’s saying. Show me the statute, show me the case, don’t just tell me. I think it’s kind of a similar approach.

Smart Guardrails for AI in the Boardroom

JR: I’m curious if you have some practical advice — this is the Practical Policies podcast, after all — for those advising on governance matters and for board members. What are some best practices they can employ in making AI enhance the boardroom experience?

ML: I would start with a conversation with your management team. If you haven’t already done this, the first question is to ask your management team: “How are you using AI in the business? I’d like to have a report and explanation of how you’re using AI in the business, what you’ve considered about it, how it’s integrated.” That’s number one. If you haven’t done that, no matter what business you’re in, you’re probably not doing what you need to do. So that’s important. Ask management those questions. “Have you talked about this with our IP attorney? Have you discussed this with our employment attorney? How are these records being kept? What are the contracts we have with our AI software providers and what are the terms? Who owns the product? Are they using our information to develop competitive technologies? How is the data being used?” Those are questions you want to ask your management team just to run the business.

As far as board members, I don’t think it’s a great idea for board members to use ChatGPT to perform their board deliberations or thought processes in AI. That’s discoverable. Let me give an example. I wouldn’t want a board member to say, “Gee, I’d really like to know how to shelter assets from this creditor.” That would be a disaster. You would have exposed the board, yourself, to personal liability or to you hiding assets. Maybe the director didn’t mean that, but you now have a discoverable chain of your thought process. The official minutes are the record of the board meeting for a reason. But I think it is acceptable to ask your service providers and the human beings that Delaware does allow you to rely upon. How are they using it? Who’s preparing your securities filing? Did they run a check to see? Did it comply with things? How does it compare to market? How are they using the technology to make those processes more efficient? It should be a conversation.

For some companies, particularly technology companies, it likely does make sense to have their own AI policy, or to at least make sure that appropriate AI use is addressed in a company policy, particularly if you have a large number of employees.

Closing Thoughts: Accountability Will Always Be Human

JR: I’m going to tie this all up with a bow and say that human oversight is always necessary in order to discharge both the duty of care and the duty of loyalty — because at the end of the day, it’s humans who are vested with these obligations.

ML: Accountability. That’s the last piece. Who are you going to blame? If you were a board member and the board said to you, “We do not believe that you exercised the duty of care, this seems like it was randomly done. Prove to us that you exercised the duty of care.” If you just turn over the chat bot, who’s accountable? Maybe you could say you would sue ChatGPT, but I’d say good luck! You have a responsibility, you know?

JR: In fact, there’s a growing body of law on that topic as well. But that’s a podcast for another day.

Melanie Levy, thank you so much for joining me today. This has been informative and interesting — and scary at some level. At the end of the day, I think this is a reality of board life now, and everybody needs to wake up to it and understand how it can enhance board performance. I think there’s something to be said for that. I also think there’s something to be said for, as you said, I’m going to quote you: “No one sits with unanswered questions.” Sometimes the thought and analysis is what gets us to the elevation of being a true human. But we’ll leave that for another podcast.

For more information about our practice, our corporate practice, employment practice and our board governance practice, visit the Insights Center at Mintz.com, or you can find us on Spotify. I’m Jen Rubin. Thanks again for joining us.

Subscribe To Viewpoints