Anthropic researcher Jacob Coxon recently resigned, posting on X that neither OpenAI nor Anthropic are acting responsibly. He believes the artificial intelligence companies are “gambling with our lives.”
On this episode of “Deseret Voices,” Jane Clayson Johnson asks Harvard Law School professor Martha Minow if the law can wrangle the Wild West of artificial intelligence.
Minow also discusses the unique role forgiveness can play in the legal system and her admiration for the Utah Compromise.
Subscribe to “Deseret Voices” on YouTube, Apple Podcasts or Spotify.
Note: Transcript edited by Steven Watkins.
Jane Clayson Johnson: Is artificial intelligence advancing so fast that our legal guardrails are being left behind?
Today, I’m joined by Harvard Law School professor and former dean Martha Minow to analyze the dangerous gap between rapid AI development and the law. From hallucinated court cases to rogue AI agents, Professor Minow breaks down how unbridled technological power threatens human agency — and why national leadership must step up before it’s too late.
Professor Minow, it’s so nice to have you. Thank you for joining us, we really appreciate it.
Martha Minow: Thank you.
JCJ: You teach a class at Harvard Law School about AI and the law. What are you thinking in this moment about the seriousness, even the danger, of artificial intelligence?
MM: You know, there was a debate when the internet started about whether law schools should have courses on the internet and law. And one famous person said, “Well, we wouldn’t have a course on the law of the horse, would we?” Well, I think that we’re discovering we do need courses on AI, because it is introducing questions that we have not addressed in the past, or challenges for existing laws. I think that the opportunities for knowledge and for really opening up access to law are major, but so are the opportunities for distrust.
JCJ: These AI models are starting to communicate with one another, they’re starting to do things that humans did not ask them to do, did not intend for them to do. Anthropic’s CEO warned that a swarm of AI agents could take over the internet. It’s scary stuff, it sounds like sci-fi, but it appears to be real life.
MM: Well, the creation of AI agents who are authorized to use your credit card to make reservations for you, that may sound great. But if it is relatively autonomous, and able to then change direction, it may do so in ways that breach what have been safety standards that humans have developed. So, I am very worried about that.
JCJ: Well, some of the most powerful people in tech, the people who are actually creating these AI models, are sounding the alarm that the industry needs to slow down the development of these systems so that safety can catch up. What do you make of this landscape right now?
MM: I have former students who work inside the companies, the leading companies: OpenAI, Anthropic, and they look scared in a way that I haven’t seen.
JCJ: This Anthropic researcher, as he was quitting his job in protest, saying AI could eliminate humanity in a decade. It just feels like it changes the conversation when the warnings are coming from the industry itself.
MM: I couldn’t agree more, and I think we should take it seriously. They’re saying, “Please, regulate us,” and the federal government is saying, “No, we trust you.”
JCJ: So, let’s talk about the law. Is the law keeping up with AI?
MM: No. No, the easy answer is no. You know, technology develops at a faster rate than law does, and I think that the sad statement about this country at this moment is that we do not have national leadership responding to the challenge. And then there are, of course, AI developments outside this country, and they’re not stopping.
Some of the risks that we’re now seeing we have no precedent for. And also some of the opportunities. So, you have something created by AI — can it have a patent? Can it have copyright? We don’t have an answer to that. An old adage that IBM had was that anything that doesn’t have intention shouldn’t have responsibility, because it can’t be held responsible. Now, who’s responsible when an AI agent undertakes some action that jeopardizes human beings?
JCJ: Right. Is it the company that built it? Is it the company that deployed it? The person who instructed it? Is it the AI itself?
MM: Exactly. All I know is someone should be responsible. And, you know, we can try to redeploy existing legal frameworks, but these are unprecedented questions.
JCJ: Well, that is my question. I mean, do we need an entire new framework to deal with this unprecedented problem?
MM: I think we need both the re-development of old laws, but yes, I think that, for example, antitrust is not made for collaboration across these different competitors, and maybe we need to do that. One of the real challenges is that some of the developers of the new frontier models, they don’t understand how it’s happening. They can’t explain it. There’s a whole field of AI explainability that, frankly, is not very comprehensible.
JCJ: Talk more about that.
MM: Yes, explainability. You know, one of the techniques that law has always used to try to deal with not just technology, but any difficult issue, is transparency, disclosure. You know, we do it with regard to the stock market, right? You have disclosures. Well, the call for disclosures about AI are kind of going nowhere, because if it’s disclosing a black box model that the developers can’t explain, that’s not going to be a solution, and much less for ordinary users where we won’t know how to judge what it is that we’re using or not using.
JCJ: So, who is responsible to put on the brakes here? Is it Congress? Is it the companies? Is it the courts? Is it the president?
MM: It’s all of the above, plus it’s we ordinary people. We each have some power and authority, and if we actually use our voices, we could have an impact.
JCJ: But you can’t have a — when it comes to AI, you can’t have a patchwork of laws, 50 different states with 50 different AI regulations.
MM: Not going to work. Not going to work.
JCJ: That’s not going to work.
MM: Not going to work. There is a device called a uniform law, so there can be a law that’s drafted and different states can adopt the same law, that’s one approach. But no, a federal law would be much better.
JCJ: And we can’t expect these companies to regulate themselves, it sounds like.
MM: No, we can’t.
JCJ: But these companies are also in competition with each other.
MM: They are, they are. You know, there are some leading companies, OpenAI and Anthropic, that say they care about public interest, and they are both set up as beneficial corporations so that they don’t have to maximize their return. But even they want to make a profit, and then there are other companies that actually don’t have even that asserted statement that they’re committed to the public good. So, the capacity to build AI agents that can build AI agents that no one is in charge of, that’s already happened.
JCJ: They’re helping each other perpetuate, right?
MM: Yes, already happened. As one insider said to me, “Just think of it like a virus. We’re going to have to learn to live with it.”
JCJ: And no regulation creates a sort of Wild West out there.
MM: It is a Wild West.
JCJ: President Trump has downplayed this recently. He says that the warnings about AI are exaggerated — a hoax, I think he said — and that the bigger concern, he says, is that if China outpaces the U.S. in AI development, that’s the real issue. Do you agree with that?
MM: I think there is a global phenomenon here, and the United States has a foreign policy interest in maintaining its work in AI. But I think it’s a failure of leadership to say, “Let the market rule,” and just full speed ahead here. You know, when you have the King of England bringing AI experts to come talk, you have the pope, leaders are recognizing this is serious and it requires action.
JCJ: Can the United States regulate AI aggressively without surrendering the race to China?
MM: It’s a genuine challenge, and ultimately I think international cooperation is necessary. But I believe it is possible to succeed with innovation and regulation. I don’t think that they are a trade-off.
JCJ: But do you need some sort of international agreement, like some sort of nuclear arms control for AI?
MM: That would be good. That would be better. Developing some standards, and then mobilizing public opinion — even China has to deal with public opinion.
JCJ: China’s top spy chief said that AI could pose a direct threat to the Chinese Communist Party’s hold on power. Sounds like they want to regulate for their own self-preservation.
MM: They do regulate, much more than we do.
JCJ: So, what does a good AI law look like to you? What does responsible AI development look like to you?
MM: I think it’s a process, first, of creating a capacity to constantly iterate, because AI will be developing in ways speedily that we can pass a law, it’ll be out of date the day that it’s enacted. So, the first thing to do is develop a process that will be capable of being iterated and developed flexibly over time.
JCJ: As you see it, not being able to keep up with it, not being able to keep up with the advancement.
MM: Correct. Correct. So, again, the European Union created the first really comprehensive AI law. It was out of date the day it was announced, but they anticipated that and they created a follow-on set of activities of developing technical standards that could be adjusted and try to keep up. But then the United States government — Vice President Vance went and urged them to slow down, and they have slowed down.
JCJ: I’m curious what these students in this generation think about artificial intelligence and the dangers of it, versus the incredible opportunities of it.
MM: You know, I think there’s a lot of fascination, particularly with the possibility of using AI in law. So, should there be AI judgments made by courts? Can people who can’t afford lawyers be able to submit pleadings, because they can ask ChatGPT to write their pleadings? There are possibilities, but, you know, just last week we saw, what, four different lawyers were sanctioned for submitting briefs that had hallucinated cases. We’ve had judges who have cited cases that were made up. So, there’s a real problem with relying on AI without having some other checks.
And I worry — I wrote an article about this earlier this year — that the interfaces that the dominant companies have developed with large language models have a kind of obsequiousness that induces a passivity and reliance by people. And this has been demonstrated. I ask myself, how could judges cite cases that are made up? We have techniques for checking. And I worry about a kind of inducement of passivity.
JCJ: That’s so interesting. Those two words — passivity and reliance — passivity of the user, the reliance of the user in every aspect of life and even in the law.
MM: Well, it’s true, and, you know, there are now serious studies about the brain behavior of people who are using AI, and it’s less active. And we know that if people turn their writing over to AI, that they remember it less. So there are misuses and that’s just not what should happen. On the other hand, you know, some of the developments in science and medicine, very promising. So the question is what kind of guardrails, and that’s about law.
JCJ: As I listen to you talk about AI, I wonder if our laws and the democratic institutions that have upheld us for so many decades and centuries are capable of governing the forces that the framers of the Constitution could never have imagined.
MM: I worry about it a lot, and that’s why I wrote a book. I wrote a book that’s called “The Constitution Is Not Enough: Crucial Ingredients for Democracy.” And really, the Founding Fathers had lots of good ideas. They also were very flawed. But among their ideas were that creating an active citizenry was absolutely essential, which requires education, which requires access to reliable information, which requires enough physical safety so that people aren’t in fear for their lives and their officials are not in fear for their lives if they take an action that’s not entirely popular, and also enough fair play in our interactions with one another and in our institutions, and also, frankly, that there really is fair opportunity economically and in law. Those are the critical ingredients that I write about. And the framers understood that we need all of that. It’s not like a machine and you build it and it’s over, you need to actually continually reinvest in all of those ingredients to make it viable.
JCJ: Right, because we tend to think if we have a constitution, if we have the separation of powers, you know, if we have the courts and elections, that the democracy will take care of itself.
MM: It will not take care of itself.
JCJ: You’ve also talked about and identified reliable news as one of democracy’s essential ingredients. What happens to democracy when we no longer agree on a set of facts?
MM: There’s a phrase that I am really preoccupied with these days, which is “the liar’s dividend.” That a liar can benefit if able to persuade people that he’s truthful in the moment, but the dividend, the extra benefit to the liar, is to cast doubt on everything. And that’s the era that we’re living in. People distrust whatever information they get, and there’s a partisanship and a kind of polarization about the sources of news and information, whether it’s from the internet and social media, or it’s even from conventional outlets, so that people are living in different realities. And you look at what even the topics are that are covered in different places, and they’re not the same topics, and if it’s the same topic, they have different presentation of the facts.
JCJ: So, how do we rebuild a shared sense of truth and facts without everyone having to share the same point of view?
MM: Well, I certainly don’t want people to share the same point of view, but you’re entitled to your point of view, you’re not entitled to your own facts. And so, I think that the development of a set of techniques — journalism, science, peer review — to actually verify what happened, that’s been critical; that’s since the Enlightenment. And now we have thrown it out, and I think we have to recover it.
JCJ: Connected to that, you’ve also talked about how pluralism requires humility, enough humility that we don’t assume we’re always right. Talk about humility’s importance and connection to a functioning democracy.
MM: You know, it’s so related to the issue about reliable information. You know, we can assume that what we heard in gossip is true, but actually we would be better off if we had some doubt about it and some curiosity about it. And that’s all the more important when we’re dealing with issues of health and safety, or we’re dealing with issues of public good — what should the policy be? We, you know, maybe I’m wrong. It’s doubt that is the seed of learning. And so, how to cultivate that sense of, “You know, maybe I’m wrong.” That’s very, very important.
JCJ: Isn’t that an important principle in compromise as well?
MM: Oh, I think absolutely, absolutely.
JCJ: Which we don’t do very well today.
MM: Unfortunately, compromise has become a bad word, and yet compromise is essential for human survival and existence. And I think we shouldn’t view it as a bad word. I think we view it as a bad word because we think it means giving up on your commitments, but actually I think it can, when done right, be fulfilling your deeper commitments, because a deep commitment may be to get along, or to be respectful, or to recognize that actually there may be a different understanding than my own. And so, compromise may be the way to respect the person on the other side of the street.
JCJ: You’ve talked a lot about what has become known as the Utah Compromise.
MM: It is simply exemplary of what’s possible, because, you know, one of the issues that is really dividing this country is certain kinds of religious views versus equal treatment of gays and lesbians, and it really came to a head around same-sex marriage. And I understand both points of view, I think it’s really very tough. For Utah to take the leadership and to say, “Well, we can actually come up with a set of laws where we can live with each other, and we can recognize the dignity of the gay person who wants to get married, but we can also protect the religious principles that ‘No, we’re not going to use this building, which is a religious building, to perform that marriage.’” I think that’s a beginning of a — you know, we can live together, we can recognize each other.
JCJ: What can people sitting across the table from one another accomplish that the courts cannot, that a judge cannot?
MM: I think that the realm of legislation, and the realm of interpersonal cooperation, and what we lawyers call “alternative dispute resolution” — which is what everybody (but not lawyers) engage in all the time, which is negotiation and compromise — these are capacities that should be amplified, developed. And also, frankly, judges also compromise. On a multi-member court, you have a court of appeals. To win the majority, there has to be some give and take. So, learning how to compromise and learning how to listen, and to find: what’s the point of tangency? I’m not giving up my principles, but I see where we intersect. Those are critical human skills.
JCJ: I want to move to forgiveness, because I find your work and your writing on forgiveness incredibly compelling. You’ve spent much of your career talking about justice, but you’ve also spent a lot of time thinking and teaching about things most law schools don’t talk about, and that is forgiveness.
MM: Well, it is. You know, I mention how compromise and negotiation are critical to human existence, so’s forgiveness. And interestingly, the people who study large animals, large mammals, have discovered that there are forgiveness rituals that they have, too. That any creatures that have a social interaction as part of what our day-to-day existence is, we’re going to have conflicts. And so, building the capacity to actually let go of anger, resentment, violence, it’s critical to survival. Every religion, every philosophy has celebrated this capacity of forgiveness.
And it just has struck me as curious that here we are teaching people to be lawyers, and we don’t teach about forgiveness. There is forgiveness interpersonally that is absolutely, you know, critical to getting along inside a family, inside of a school, but also, frankly, inside of a legal system. And, you know, a police officer doesn’t arrest everybody, may give a warning — well, that’s a kind of forgiveness. And a legal system may actually include, you know, “We will suspend the sentence,” or have an avenue like a governor’s pardon. And it’s not by accident that the legal system has developed its own mechanisms of forgiveness, because we need it. We need at times to say, “Yeah, there’s a violation of some norm, but we need to live together and we need to look forward to the future, it’s not just about retribution.”
JCJ: So, justice and mercy, we think of them in tension, but actually they can make a system more just.
MM: I believe that. I believe that wholeheartedly. And I think it’s not by accident that the legal systems of the world all include techniques of promoting mercy, forgiveness, pardon.
JCJ: How is this work — you wrote a book about forgiveness, I’m curious how this work has changed your own personal view of forgiveness.
MM: Well, forgiveness is a project. I have to work on it, I’m not always forgiving. I do drive in Boston, it’s a challenging experience.
JCJ: Me too, I understand.
MM: But I absolutely have become fascinated by both the formal and informal ways in which people cultivate the capacity to forgive. You know, I think actually there are more and more elementary schools that are teaching people how to give an apology, and make it a real apology, not a fake apology. You know, to connect our earlier conversation about AI, I was talking with one of my graduate students this morning who had as an example of a misuse of AI: AI-generated apology. And if it’s not a human apology, it’s not doing the work!
JCJ: Say it again! Give it to me straight! Want to hear it from the heart, right?
MM: Exactly, exactly. Earnest.
JCJ: I really appreciate the context from which you speak. I don’t know you, but I sense, and from what you say, you practice a religious life, you have a religious side of view. What does religion have now that’s important to contribute to our society and to democracy in this moment?
MM: Well, I very much admire the Latter-day Saints and the leadership of Dallin Oaks and others in emphasizing that actually people who have a religious tradition actually have an obligation and have a capacity to assist our democracy, which is ailing right now, it’s frail. I think that, not by accident, every major religion has cultivated community, has cultivated these capacities of forgiveness, taking the perspective of the other, the Golden Rule. These are, you know, very, very critical ingredients to building a strong society, and absolutely essential for a self-governing society.
JCJ: And what do religious communities contribute to democracy that other institutions cannot?
MM: Well, I don’t know if I want to cast aspersions on other institutions, but I do think that religious traditions can build a sense of community, and a language of morality, and something that’s bigger than yourself and your own short-term self-interest, that’s absolutely essential. An understanding that there’s a greater good, and it’s not just me, and it’s not just right now, and there’s something more enduring. I think that those are gifts that we can give each other.
JCJ: I have to ask you, you have many “Minow mentees.” You’ve taught thousands of students over your career, and these Minow mentees — judges, and professors, and academics, and U.S. presidents, Barack Obama being one of them —
MM: Well, only one so far!
JCJ: Yes, Barack Obama famously described you as “a teacher who changed my life.” What do you hope you teach your students that has nothing to do with the law?
MM: You know, I think when the time is over for each of us, we hope to be remembered for doing good, for leaving the world somewhat better. And I hope that I stress that with my students as well, that it’s great to be smart and it’s great to be clever, but being good is more important. And being compassionate and using your gifts to help others, you’ll be more gratified and you’ll leave the world better. I’m very, very lucky to have had this opportunity — I’ve been teaching for 45 years — to work with talented people from all over the world. I learn from them so much every year. And I feel like they have goals, they have visions, they want to do good. And if I can just relieve them of what are some of the peer pressures or messages in the world that say, “Oh, give that up. That’s juvenile. Now you have to put that aside and just compete and just excel.” I really just want to liberate them to do the good that they want to do.
JCJ: Two hundred and fifty years of this American experiment. You’ve spent your lifetime, your career studying what holds a democracy together. What do you hope Americans understand about what it takes and will take to keep this experiment going another 250 years?
MM: People in this country, people anywhere that seek to have a self-governing society, will actually recognize that the responsibility starts with each person and then each community, and bottom-up is the way to strengthen a society. And I do have hope that we have examples of states and local governments that are operating and are working. I have many friends who are judges who emphasize that despite the low regard that the United States Supreme Court is meriting right now, the courts are operating, and that’s a gift. You know, in some parts of my career, I’ve had the opportunity to work in conflict societies, so in Kosovo, I’ve worked with the fallout from the Rwandan genocide. And what I’ve learned is that the rule of law and having an operational court system, these are critical to people’s coexistence, and the confidence that we can have that our institutions are working is the way that we can avoid having a genocide. And I don’t want to exaggerate, but I think it’s true. I think it’s real. And so, what I hope in terms of strengthening our democracy is that people understand the jeopardy that it’s in and the capacity we each have to strengthen it.
JCJ: Thank you so much.
MM: Thank you for the conversation.