The comprehensive, even sweeping, set of guidelines^[1] for artificial intelligence that the White House unveiled in an executive order on Oct. 30, 2023, show that the U.S. government is attempting to address the risks posed by AI.

As a researcher of information systems and responsible AI^[2], I believe the executive order represents an important step in building responsible^[3] and trustworthy^[4] AI.

The order is only a step, however, and it leaves unresolved the issue of comprehensive data privacy legislation. Without such laws, people are at greater risk of AI systems revealing sensitive or confidential information^[5].

Understanding AI risks

Technology is typically evaluated for performance, cost and quality^[6], but often not equity, fairness and transparency. In response, researchers and practitioners of responsible AI have been advocating for:

• algorithm auditing^[7]
• standard reports on AI models^[8]
• credentials for otherwise opaque AI systems^[9]
• comprehensive risk mitigation practices^[10]
• AIs that are transparent to the public^[11]
• a recognition of the harms caused by AIs^[12] that make predictions about people

The National Institute of Standards and Technology (NIST) issued a comprehensive AI risk management framework^[13] in January 2023 that aims to address many of these issues. The framework serves as the foundation^[14] for much of the Biden administration’s executive order. The executive order also empowers the Department of Commerce^[15], NIST’s home in the federal government, to play a key role in implementing the proposed directives.

Researchers of AI ethics have long cautioned that stronger auditing of AI systems^[16] is needed to avoid giving the appearance of scrutiny without genuine accountability^[17]. As it stands, a recent study looking at public disclosures from companies found that claims of AI ethics practices outpace actual AI ethics initiatives^[18]. The executive order could help by specifying avenues for enforcing accountability.

Another important initiative outlined in the executive order is probing for vulnerabilities of very large-scale general-purpose AI models^[19] trained on massive amounts of data, such as the models that power OpenAI’s ChatGPT or DALL-E. The order requires companies that build large AI systems with the potential to affect national security, public health or the economy to perform red teaming^[20] and report the results to the government. Red teaming is using manual or automated methods to attempt to force an AI model to produce harmful output^[21] – for example, make offensive or dangerous statements like advice on how to sell drugs.

Reporting to the government is important given that a recent study found most of the companies that make these large-scale AI systems lacking^[22] when it comes to transparency.

Similarly, the public is at risk of being fooled by AI-generated content. To address this, the executive order directs the Department of Commerce to develop guidance for labeling AI-generated content^[23]. Federal agencies will be required to use AI watermarking^[24] – technology that marks content as AI-generated to reduce fraud and misinformation – though it’s not required for the private sector.

The executive order also recognizes that AI systems can pose unacceptable risks^[25] of harm to civil and human rights^[26] and the well-being of individuals: “Artificial Intelligence systems deployed irresponsibly have reproduced and intensified existing inequities, caused new types of harmful discrimination, and exacerbated online and physical harms.”

What the executive order doesn’t do

A key challenge for AI regulation is the absence of comprehensive federal data protection and privacy legislation. The executive order only calls on Congress to adopt privacy legislation, but it does not provide a legislative framework. It remains to be seen how the courts will interpret the executive order’s directives in light of existing consumer privacy and data rights statutes.

Without strong data privacy laws in the U.S. as other countries have, the executive order could have minimal effect on getting AI companies to boost data privacy. In general, it’s difficult to measure the impact that decision-making AI systems have on data privacy and freedoms^[27].

It’s also worth noting that algorithmic transparency is not a panacea. For example, the European Union’s General Data Protection Regulation legislation mandates “meaningful information about the logic involved^[28]” in automated decisions. This suggests a right to an explanation of the criteria that algorithms use in their decision-making. The mandate treats the process of algorithmic decision-making as something akin to a recipe book, meaning it assumes that if people understand how algorithmic decision-making works, they can understand how the system affects them^[29]. But knowing how an AI system works doesn’t necessarily tell you why it made a particular decision^[30].

With algorithmic decision-making becoming pervasive, the White House executive order and the international summit on AI safety^[31] highlight that lawmakers are beginning to understand the importance of AI regulation, even if comprehensive legislation is lacking.

The cars, cellphones, computers and televisions that people in the U.S. use every day require metals like copper, cobalt and platinum to build. Demand from the electronics industry for these metals is only rising, and companies are constantly searching for new places on Earth to mine them.

Scientists estimate that lots of these metals exist thousands of miles beneath Earth’s surface, in its molten core, but that’s far too deep and hot to mine^[1]. Instead, some companies hope to one day search for deposits that are literally out of this world — on asteroids.

The commercialization of asteroid mining is still a ways off, but in October 2023, NASA launched a scientific mission to explore the metal-rich asteroid Psyche^[2]. The main goal of the mission^[3] is studying the composition and structure of this asteroid, which could tell scientists more about Earth’s core since the two objects might have a similar makeup.

Both likely contain platinum, nickel, iron and possibly even gold – materials of commercial interest.

I am a planetary geologist^[4] whose work explores other planets and astronomical objects like Mars, Venus and the Moon. I will be following the Psyche mission closely, as this is the first time that scientists will be able to learn about the composition and structure of a possible piece of a planetary core similar to the Earth’s^[5], without indirect seismic or magnetic measurements, or replicating the pressure and temperature conditions of the Earth’s core in our labs.

With the spacecraft estimated to arrive at the asteroid’s orbit in 2029, the findings from the Psyche mission will provide unique insights into the type of metals present on the asteroid’s surface, as well as their amount, and the minerals containing these metals. This data is essential both for scientists like me exploring the formation and evolution planetary bodies, as well as for companies investigating the possibility of asteroid mining.

Asteroid formation

Asteroids come in a variety of sizes^[6]. Some are the size of a town, while others are the size of a state. Most asteroids are made of rocks and represent the leftovers from the early formation of our solar system^[7] around 4.6 billion years ago.

Not every asteroid is the same – some, like Bennu, the target of NASA’s OSIRIS-REx mission^[9], are rich in carbon. These are very old, and they will teach scientists more about how planets formed and how life may have begun on Earth.

Others, like Psyche^[10], are made of metals and potentially result from one or more collisions between astronomical objects when the solar system was forming. These collisions left debris flying through space — including potential pieces of a planet’s metal-rich core. A NASA spacecraft will orbit and analyze the surface of Psyche.

Mining in space

Not every mineral deposit on Earth is mineable. Companies first look for deposits with a high level of metal purity^[11]. They also investigate how affordable and feasible extracting the metal would be before choosing where to mine.

Similarly, before mining an asteroid, companies will have to think about all those factors, and they’ll have to come up with the infrastructure needed to mine at a distance and transport the metals they mine hundreds of millions of miles back to Earth. The technology to do that is still years away, and transporting metals would require major funding.

A few companies^[13] around the world have already started to think about what the best and lowest cost approach would be, drawing from processes similar to those used on Earth.

The first step would be finding a mineable metal deposit^[14]. Next, they’d drill and extract the metals on the asteroid^[15]. One of the most important differences with Earth mines is that each step would be undertaken remotely with spacecrafts orbiting around the asteroid and robots landing on its surface. Then, a spacecraft would send the resulting materials back to Earth.

Asteroid mining plans are still at their earliest stages. A few companies like Planetary Resources^[16] and Deep Space Industries^[17], with goals to extract metals from space, were acquired by other companies.

Experts can’t quite tell yet how acquiring valuable metals from asteroids would affect the global economy, but these metals could potentially flood the market and lower their values^[18].

The Psyche mission is a huge step in figuring out what sort of metals are out there, and it may also answer questions about the composition and properties of Earth’s core.

The benefits of friendship go far beyond having someone to confide in or spend time with – it can also protect you from physical and mental health problems. For example, people with good friends recover more quickly from illnesses^[1] and surgeries^[2]. They report higher well-being^[3] and feel like they live up to their full potential^[4]. Additionally, people with good friends report being less lonely across many life stages, including adolescence^[5], becoming a parent^[6] and old age^[7].

In fact, friendships are so powerful that the social pain of rejection activates the same neural pathways that physical pain^[8] does.

Behavioral scientists like me^[9] have tended to focus our research about friendships on their benefits. How to cultivate these powerful relationships hasn’t been as deeply researched yet. Understanding more about what people look for in a friend and how to make and sustain good friendships could help fight the loneliness epidemic^[10].

Traditional conceptions of friendship

Previous generations of behavioral scientists traditionally focused on the notion that people form friendships with those who are similar^[11], familiar^[12] and in close proximity^[13] to them.

When you look at all the friendships you’ve had over your life, these three factors probably make intuitive sense. You’re more likely to have things in common with your friends than not. You feel an increased sense of familiarity with friends the longer you know them – what psychologists call the mere exposure effect^[14]. And your friends are more likely to live or work near you.

Researchers in this field have also typically divided friendship preferences based on gender. The dichotomy suggests that women prefer one-on-one^[15], emotionally close^[16] and face-to-face^[17] friendships, while men prefer multi-person, task-oriented and side-by-side^[18] friendships, with the focus on a shared activity.

Again, when looking at your own friendships, these findings may seem intuitive. Women on average prefer to engage in activities that allow for self-disclosure and sharing secrets, such as spending time one-on-one talking about their lives. Men, on the other hand, tend to prefer to engage in activities that are group-based and have a clearly defined outcome, such as playing sports together. Findings such as these show that gender^[20] and preferences on how to connect^[21] are important in friendships.

But these explanations of friendship do not address the most important aspect of making friends – choosing the individual people you want to turn into your pals. Friendship decisions are not random. There are many people who are similar, familiar, in close proximity and have similar preferences as you. Yet few of these individuals end up being your friends.

So, in a world full of possibilities, how do people pick those who will become their friends?

New ways to think about friendship

Within the last decade, researchers have begun investigating the roots of friendship preferences beyond the classic descriptions.

For example, social scientists see there are strong preferences for friends to be loyal, trustworthy^[22] and warm^[23]. Additionally, researchers find there are preferences for friends who help you solve specific kinds of problems^[24] and are generous and caring with you^[25] instead of others. These preferences help people navigate making friends, given limited reserves of time and effort^[26]. In short, they help you find the best possible friends you can in a world full of friendship possibilities.

Social scientists have also learned that, while there are some important gender differences in what people want in friends, it is not accurate to say that men and women want one kind of friendship over another. In fact, when we take a more holistic approach and consider^[27] broader categorizations of^[28] emotional closeness^[29] and tasks^[30], the gender differences in these preferences are reduced. And of course, people don’t exclusively pick between face-to-face and side-by-side friendships. Instead, it is more likely that they focus on what they want from their friends and let these needs guide how friendships form.

Ultimately it’s your individual preferences that guide you toward the people who will best meet your particular social needs. With a little luck, you’ll find buddies who can lend a hand when you need one and support you in reaching your goals. In all, your preferences are the key to finding friends who can buffer against feeling lonely and provide you with the social, emotional and health benefits of friendship.

When you’re looking for friends

It’s hard to provide clear guidelines for improving friendships because the research about friendship preferences is still developing. But there are some clear points for consideration:

1. Determine what you value in friends. Do you want one-on-one, emotionally close friendships or multi-person, task-oriented friendships? Depending on your preference, different kinds of activities will be helpful for finding others who fit the bill and cultivating these friendships.

2. Know that it will take time to make close friendships^[32]. Research suggests that it takes 30 hours of interaction to make a casual friend, 140 hours to make a good friend and 300 hours to make a best friend.

3. Consider what you bring to the table. Everyone has unique strengths they bring to their friendships. Research shows that, when you’re able to demonstrate that you have characteristics people want in friends, you’re able to make more satisfying friendships^[33].

Understand friendships to understand loneliness

Considering the nuances of friendship preferences will be extremely important in reducing not only loneliness, but other related public health crises. For example, loneliness is associated with likelihood of attempting suicide^[34]. Recent surveys have found that men are suffering big declines in the number of close friends^[35] they have, as well as experiencing higher rates of suicide^[36] compared to women.

The U.S. Surgeon General’s recent recommendations for fighting the loneliness epidemic focus on public policies and infrastructure^[37]. But fostering community spaces for connection – such as parks, libraries and playgrounds – prioritizes the preferences of those who favor the one-on-one, emotionally close and face-to-face connections more often preferred by women. These places are less beneficial for people with more typically masculine preferences, as there is no guarantee that these spaces will foster side-by-side, task-oriented connections unless areas for sports and other team-based activities are also included.

To counter this inequity, researchers and public health officials first need to understand what makes friendships satisfying. Then they can ensure that recommendations to curb loneliness address all of the pathways that people use to cultivate high-quality friendships.

How researchers conceptualize a disease informs how they treat it. Cancer is often described as uncontrollable cell growth triggered by genetic damage. But cancer can also be seen from angles that emphasize mathematics, evolutionary game theory and physics, among others.

Molecular biology has brought significant advances in making it possible to live with cancer as a chronic illness rather than a fatal disease. Alternative frameworks, however, can offer scientists additional insights on how to prevent tumors from spreading throughout the body and becoming resistant to treatment.

Here are a few unconventional lenses through which researchers are viewing cancer with fresh eyes, drawn from The Conversation’s archives.

1. Evolution and natural selection of cancer

The body is far from a wonderland for cells. Each individual cell competes against trillions of others for finite space and nutrients. If they’re able to cooperate in an orderly enough fashion, sharing resources and dividing labor, the collective functions effectively. Cancer cells, however, cheat the system^[1]: They hog resources, take up as much space as possible and refuse to die^[2].

In this way, cancer can be thought of as an evolutionary disease^[3] – these are cells that have developed the genetic mutations to outcompete their neighbors, and subsequent cell generations inherit this survival advantage. Cancer cells benefit at the expense of the collective until the entire organism collapses.

Oncologist Monika Joshi^[5] and pathologists Joshua Warrick^[6] and David DeGraff^[7] believe that understanding evolution is key to understanding cancer. Screening programs are effective, for example, because removing a nascent tumor is easier than treating one that has evolved the ability to spread. Cancer cells likewise become resistant to treatments because they’re pushed to further evolve to survive.

Some researchers are applying the principles of evolutionary game theory to reduce treatment resistance^[8] and optimize therapies for children^[9].

“The fight against cancer is a fight against evolution, the fundamental process that has driven life on Earth since time immemorial,” they wrote. “This is not an easy fight, but medicine has made tremendous progress.”

---

Read more: Every cancer is unique – why different cancers require different treatments, and how evolution drives drug resistance^[10]

---

2. Fluid mechanics of cancer

As much as cancer is a disease that respects no boundaries, tumor cells are still shaped by their environment. Unlike healthy cells that take the hint when their presence isn’t wanted, however, tumor cells not only survive but thrive in stressful conditions^[11]. Isolated cancer cells able to adapt to harsh settings are the ones that establish metastatic colonies and become resistant to treatment.

While researchers have focused on how biochemical signals direct cells to move from one location to another, a cell’s physical environment also affects where it migrates. Mechanical engineer Yizeng Li^[12] found that a cell’s “solid” and “fluid” surroundings influence its movement.

Cancer cells encounter varying degrees of fluid viscosity, or thickness, as they travel through the body. Li and her team found that breast cancer cells counterintuitively move faster in high viscosity environments by changing their structure. This meant that fluid viscosity serves as a mechanobiological cue for cancer cells to metastasize^[13].

“Cancer patients usually don’t die from the original source of the tumor but from its spread to other parts of the body,” Li wrote. “Understanding how fluid viscosity affects the movement of tumor cells could help researchers figure out ways to better treat and detect cancer before it metastasizes.”

---

Read more: How cancer cells move and metastasize is influenced by the fluids surrounding them – understanding how tumors migrate can help stop their spread^[16]

---

3. Inflammation link to cardiovascular disease

Apart from being leading causes of death around the world, cardiovascular disease and cancer may not initially seem to have much in common. The many risk factors they share, however – like poor diet, smoking and chronic stress – coalesce with chronic inflammation: persistent, low-grade activation of the immune system can damage cells in ways that encourage either disease to develop.

For biomedical engineer Bryan Smith^[17], the developmental parallels between these diseases signal they could be treated at the same time^[18].

Drugs can be repurposed^[19] to target diseases for which they weren’t originally designed. Certain drugs, for example, can direct immune cells called macrophages to consume both cancer cells and the cellular debris that contribute to cardiovascular plaques.

“As basic science discovers other molecular parallels between these diseases, patients will be the beneficiaries of better therapies that can treat both,” wrote Smith.

---

Read more: Could a single drug treat the two leading causes of death in the US: cancer and cardiovascular disease?^[20]

---

4. Mathematics of cancer

In certain contexts, math has unique strengths in describing the natural world^[21]. For instance, epigenetics – where and when genes are turned on or off – plays as much a role in cancer progression as direct changes to the genetic code. Epigenetic changes can alter healthy cells to the point of losing their normal form and function. But the randomness of these changes makes it difficult to tease out pathological from normal genetic activity.

A mathematical concept called stochasticity – or how the randomness of the steps of a process influences how predictable its outcome will be – lends a logical framework to the epigenetic changes contributing to cancer^[22], clarifying when healthy cells rapidly develop into tumor cells.

Stochasticity is commonly used to study stock market behavior and epidemic disease spread, and researchers quantify it by examining the degree of uncertainty, or entropy, of a particular outcome. Identifying high entropy areas in the genome could offer another approach to cancer detection and drug design.

Cancer geneticist Andrew Feinberg^[23] has been using entropy to quantitatively describe the epigenetics of cancer. He and his colleagues found that high entropy regions of the genome in the skin become even more entropic with sun damage, increasing the chance of developing cancer. This offers a potential explanation for why cancer risk significantly increases with age.

“Epigenetic entropy shows that you can’t fully understand cancer without mathematics,” Feinberg wrote. “Biology is catching up with other hard sciences in incorporating mathematical methods with biological experimentation.”

---

Read more: Cancer evolution is mathematical – how random processes and epigenetics can explain why tumor cells shape-shift, metastasize and resist treatments^[24]

---

5. A public health issue

Cancer is a disease that develops in an individual, but its socially derived causes and societal-wide effects are hardly limited to a single person.

Take the case of lung cancer. It is stigmatized as a disease brought on by poor lifestyle choices – a consequence of a personal decision to use tobacco products. But as thoracic oncologist Estelamari Rodriguez^[25] noted, the face of lung cancer has changed.

“Over the past 15 years, more women, never-smokers and younger people are being diagnosed with lung cancer,” she wrote. While lung cancer rates have significantly decreased for men, they have substantially risen for women^[26] around the world. Despite being the leading cause of cancer death among women, screening rates remain low compared with other cancers.

More broadly, cancer symptoms are often unrecognized or misdiagnosed, not only for women^[27] but also for many marginalized populations, including people of color^[28], transgender patients^[29] and the uninsured^[30].

These disparities are due in part to biases in medical education and clinical research^[31] that fail to prepare clinicians to care for the diversity of patients they’ll encounter. Tenuous access to preventive care^[32] and disproportionate exposure to carcinogens^[33] among certain populations compound these inequities.

The purview of cancer goes far beyond a single discipline. It takes a village of researchers, policymakers and patient advocates to achieve effective and accessible cancer care for all.

---

Read more: Lung cancer rates have decreased for the Marlboro Man, but have risen steeply for nonsmokers and young women – an oncologist explains why^[34]

---

For centuries, the quest for new elements^[1] was a driving force in many scientific disciplines. Understanding an atom’s structure and the development of nuclear science allowed scientists to accomplish the old goal of alchemists^[2] – turning one element into another^[3].

Over the past few decades, scientists in the United States^[4], Germany^[5] and Russia^[6] have figured out how to use special tools to combine two atomic nuclei^[7] and create new, superheavy elements^[8].

These heavy elements usually aren’t stable. Heavier elements have more protons^[11], or positively charged particles in the nucleus; some that scientists have created have up to 118^[12]. With that many protons, the electromagnetic repulsive forces between protons in the atomic nuclei overwhelm the attractive nuclear force that keeps the nucleus together.

Scientists have predicted for a long time^[13] that elements with around 164 protons could have a relatively long half-life^[14], or even be stable. They call this the “island of stability^[15]” – here, the attractive nuclear force is strong enough to balance out any electromagnetic repulsion.

Since heavy elements are difficult to make in the lab, physicists like me^[17] have been looking for them elements everywhere, even beyond the Earth^[18]. To narrow down the search, we need to know what sort of natural processes could produce these elements. We also need to know what properties they have, like their mass densities.

Calculating density

From the outset, my team wanted to figure out the mass density of these superheavy elements. This property could tell us more about how the atomic nuclei of these elements behave. And once we had an idea about their density, we could get a better sense of where these elements might be hiding.

To figure out the mass density and other chemical properties^[19] of these elements, my research team used a model that represents an atom of each of these heavy elements as a single, charged cloud. This model works well for large atoms, particularly metals that are laid out in a lattice structure.

We first applied this model^[20] to atoms with known densities and calculated their chemical properties. Once we knew it worked, we used the model to calculate the density of elements with 164 protons, and other elements in this island of stability.

Based on our calculations, we expect stable metals with atomic numbers around 164 to have densities between 36 to 68 g/cm³ (21 to 39 oz/in³). However, in our calculations, we used a conservative assumption about the mass of atomic nuclei. It’s possible that the actual range is up to 40% higher.

Asteroids and heavy elements

Many scientists believe that gold^[21] and other heavy metals were deposited on Earth’s surface after asteroids collided with the planet^[22].

The same thing could have happened with these superheavy elements, but super mass dense heavy elements sink into ground and are eliminated from near the Earth’s surface by the subduction of tectonic plates^[23]. However, while researchers might not find superheavy elements on Earth’s surface, they could still be in asteroids like the ones that might have brought them to this planet.

Scientists have estimated that some asteroids have mass densities greater than that of osmium^[24] (22.59 g/cm³, 13.06 oz/in³), the densest element found on Earth.

The largest of these objects is asteroid 33, which is nicknamed Polyhymnia^[25] and has a calculated density of 75.3 g/cm³ (43.5 oz/in³). But this density might not be quite right, since it’s quite difficult to measure the mass and volume of far-away asteroids.

Polyhymnia isn’t the only dense asteroid out there. In fact, there’s a whole class of superheavy objects, including asteroids, which could contain these superheavy elements. Some time ago, I introduced the name Compact Ultradense Objects, or CUDOs^[26], for this class.

In a study published in October 2023 in the European Physical Journal Plus^[27], my team suggested some of the CUDOs orbiting in the solar system might still contain some of these dense, heavy elements^[28] in their cores. Their surfaces would have accumulated normal matter over time and would appear normal to a distant observer.

So how are these heavy elements produced^[29]? Some extreme astronomical events, like double star mergers^[30] could be hot and dense enough to produce stable superheavy elements.

Some of the superheavy material could then remain on board asteroids created in these events. They could stay packed in these asteroids, which orbit the solar system for billions of years.

Looking to the future

The Eurpoean Space Agency’s Gaia mission^[31] aims to create the largest, most precise three-dimensional map of everything in the sky. Researchers could use these extremely precise results to study the motion of asteroids^[32] and figure out which ones might have an unusually large density.

Space missions are being conducted to collect material from the surfaces of asteroids and analyze them back on Earth. Both NASA and the Japanese state space agency JAXA^[33] have targeted low density near-Earth asteroids with success. Just this month, NASA’s OSIRIS-REx^[34] mission brought back a sample. Though the sample analysis is just getting started, there is a very small chance it could harbor dust containing superheavy elements accumulated over billions of years.

One mass-dense dust and rock sample brought back to Earth would be enough. NASA’s Psyche mission^[36], which launched in October 2023, will fly to and sample a metal-rich asteroid^[37] with a greater chance of harboring superheavy elements. More asteroid missions like this will help scientists better understand the properties of asteroids orbiting in the solar system.

Learning more about asteroids and exploring potential sources of superheavy elements will help scientists continue the century-spanning quest to characterize the matter that makes up the universe and better understand how objects in the solar system formed.

Evan LaForge, an undergraduate student studying physics and mathematics, is the lead author on this research^[38] and helped with the writing of this article, along with Will Price, a physics graduate student.

Only “persons” can engage with the legal system – for example, by signing contracts or filing lawsuits. There are two main categories of persons^[1]: humans, termed “natural persons,” and creations of the law, termed “artificial persons.” These include corporations, nonprofit organizations and limited liability companies^[2] (LLCs).

Up to now, artificial persons have served the purpose of helping humans achieve certain goals. For example, people can pool assets in a corporation and limit their liability vis-à-vis customers or other persons who interact with the corporation. But a new type of artificial person is poised to enter the scene – artificial intelligence systems, and they won’t necessarily serve human interests.

As scholars^[3] who study AI and law^[4] we believe that this moment presents a significant challenge to the legal system: how to regulate AI within existing legal frameworks to reduce undesirable behaviors, and how to assign legal responsibility for autonomous actions of AIs.

One solution is teaching AIs to be law-abiding entities^[5].

This is far from a philosophical question. The laws governing LLCs in several U.S. states^[6] do not require that humans oversee the operations of an LLC. In fact, in some states it is possible to have an LLC with no human owner^[7], or “member” – for example, in cases where all of the partners have died. Though legislators probably weren’t thinking of AI when they crafted the LLC laws, the possibility for zero-member LLCs opens the door to creating LLCs operated by AIs.

Many functions inside small and large companies have already been delegated to AI in part, including financial operations, human resources^[8] and network management, to name just three. AIs can now perform many tasks as well as humans do. For example, AIs can read medical X-rays^[9] and do other medical tasks, and carry out tasks that require legal reasoning^[10]. This process is likely to accelerate due to innovation and economic interests.

A different kind of person

Humans have occasionally included nonhuman entities like animals^[11], lakes^[12] and rivers^[13], as well as corporations^[14], as legal subjects. Though in some cases these entities can be held liable for their actions, the law only allows humans to fully participate in the legal system.

One major barrier to full access to the legal system by nonhuman entities has been the role of language^[15] as a uniquely human invention and a vital element in the legal system. Language enables humans to understand norms and institutions that constitute the legal framework. But humans are no longer the only entities using human language.

The recent development^[16] of AI’s ability to understand human language^[17] unlocks its potential to interact with the legal system. AI has demonstrated proficiency in various legal tasks, such as tax law advice^[18], lobbying^[19], contract drafting and legal reasoning^[20].

An LLC established in a jurisdiction that allows it to operate without human members could trade in digital currencies^[22] settled on blockchains^[23], allowing the AI running the LLC to operate autonomously and in a decentralized manner that makes it challenging to regulate. Under a legal principle known as the internal affairs doctrine^[24], even if only one U.S. state allowed AI-operated LLCs, that entity could operate nationwide – and possibly worldwide. This is because courts look to the law of the state of incorporation for rules governing the internal affairs of a corporate entity.

We believe the best path forward, therefore, is aligning AI with existing laws, instead of creating a separate set of rules for AI. Additional law can be layered on top for artificial agents^[25], but AI should be subject to at least all the laws a human is subject to.

Building the law into AI

We suggest a research direction of integrating law into AI agents^[26] to help ensure adherence to legal standards^[27]. Researchers could train AI systems to learn methods for internalizing the spirit of the law^[28]. The training would use data generated by legal processes and tools of law, including methods of lawmaking, statutory interpretation, contract drafting, applications of legal standards and legal reasoning.

In addition to embedding law into AI agents, researchers can develop AI compliance agents – AIs designed to help an organization automatically follow the law. These specialized AI systems would provide third-party legal guardrails.

Researchers can develop better AI legal compliance by fine-tuning large language models with supervised learning^[29] on labeled legal task completions. Another approach is reinforcement learning^[30], which uses feedback to tell an AI if it’s doing a good or bad job – in this case, attorneys interacting with language models. And legal experts could design prompting schemes – ways of interacting with a language model – to elicit better responses from language models that are more consistent with legal standards.

Law-abiding (artificial) business owners

If an LLC were operated by an AI, it would have to obey the law like any other LLC, and courts could order it to pay damages, or stop doing something by issuing an injunction. An AI tasked with operating the LLC and, among other things, maintaining proper business insurance would have an incentive to understand applicable laws and comply. Having minimum business liability insurance policies is a standard requirement that most businesses impose on one another to engage in commercial relationships.

The incentives to establish AI-operated LLCs are there. Fortunately, we believe it is possible and desirable to do the work to embed the law – what has until now been human law – into AI, and AI-powered automated compliance guardrails.