Why both groups of advocates should celebrate this.

In this post:

• It’s no coincidence that one community was ahead of the curve in seeking to address factory farming, the danger of AI, and the potential suffering of digital minds; consistent principles point to all three.

• These movements may have needed some time to develop apart, but it’s time for them to reunite. The question of how AI takes off has become the decisive factor determining animal welfare in the mid-term future.

• Animal welfare advocates should start thinking of ourselves as part of the “Make AI Go Well” movement, because that’s where most of our leverage lies.

• Conversely, AI safety and alignment researchers should consider animal welfare to be one important part of their mandate. You alone have the skills to address this particular alignment problem, and we can’t be nearly confident enough that it will be solved by default if human alignment is solved.

Prelude

A worldwide universal income worth one million 2026 dollars per person, with one of your personal robots preparing a nutritionally-optimized, non-carcinogenic cultivated steak exactly to your liking, before you even realize you wanted one.

300 million Earth-like planets in the Milky Way terraformed, dotted with factory farms amidst vast forests where wild animals struggle and die in droves, all to satisfy humans’ “naturalistic” sensibilities.

Cities of trillions of happy people leading immortal lives of infinite luxury, eating nothing but sunlight in a server the size of a shipping container.

Earth reduced to a smoldering rock, the aftermath of an apocalyptic war between man and machine.

These are just some of the ways the world could be transformed as early as 2040– at least according to many very smart people who’ve been paying close attention to AI since long before ChatGPT.

They’ve convinced me. I don’t have much money to invest but I’ve bet it all on AI-exposed equities because that’s the closest I can come to putting my money where my mouth is. I believe any one of the above outcomes is more likely than the world of 2040 looking much like it looks now. For the world to be recognizable to us in 2100 is, I think, fundamentally out of the question.

If you can’t understand why I’d believe these things, but you’ve got some curiosity, check out this post. If you’ve already engaged with the arguments and still don’t buy it, that’s OK too. I think you’ll still find this post useful, and I promise I won’t spend it trying to convince you because that’s what that post was for.

Our focus today is: if there’s a chance that AI will transform the world this dramatically, this soon, what does that mean for the animal movement, the AI safety movement, and the relationship between the two?

Part 1: It’s No Coincidence

1.1 Family dinners

If you think these views about AI are weird, just imagine what my poor parents think.

Around ten years ago, I showed up to Christmas dinner saying I wouldn’t participate in holidays anymore unless the meals were free of animal products, because I believed that the factory farms those products came from were a moral atrocity on par with slavery and the holocaust.

They found that view pretty weird, though to their massive credit, they agreed to go along with it. Now they’ve had a decade to get used to it.

Then a couple years ago I started hitting them with the AI stuff, everything from “AI might rise up against us and kill all humans, like the terminator except it wouldn’t leave any of us alive and yeah that’s literally the default trajectory” to “digital minds, possibly even those same AI that kill all of us, might suffer so much that it makes all of factory farming look like a stubbed toe.”

Bless their hearts, my parents tried to understand. But this started to get a bit too weird for them.

But reader, the weirdest thing of all, the thing I have the hardest time explaining, is why there is so much overlap between people who hold these two sets of views. When they ask how I got so quickly from ChatGPT being released to “AI is now your highest fatality risk,” I mumble something about it being a weird historical accident that some people in the 2010s started worrying about both factory farming and AI, exposing me to AI risk ideas earlier than most people. At some point, I stopped saying the words effective altruism because it just introduced one more weird thing to explain.

1.2 First impressions

But reader, you know deep down it’s not a historical accident. It may not seem obvious at first glance, but there are good reasons that concern for animal welfare and AI safety took root in the same community– namely, Effective Altruism.

I first heard about EA in 2015, at the same time that I was getting involved in the animal rights movement. I gravitated towards the grassroots/radical/abolitionist wing of the movement, which was opposite the EA-influenced side. My tribal brain kicked in and for many years, I regarded EA with overt hostility.

Of course, I had better explanations at the time. EA was the wrong tool for animal advocacy. It had a positive impact in addressing global poverty, but animal rights was different: the powers that be were in support of poverty alleviation, but animal rights was a revolution against the status quo, and required a different type of thinking. In particular, EA’s obsession with measurability led to a self-defeating bias where they were only willing to pursue strategies with easily measurable results. My preferred approach, based on imitating past social movements like civil rights and women’s suffrage, was hard to measure in the short term. That meant we were left out in the cold as more and more animal advocacy funding became dominated by EA thinking.

You can imagine what I thought when I learned about the strain of EA dedicated to the absolutely harebrained ideas that nature was bad for wildlife and should be destroyed, and that non-player characters in video games might be capable of morally relevant suffering, making video games a moral atrocity. The fact that EAs were wasting time worrying about the wild animals who would suffer if we terraformed mars, or the software runtimes that might suffer in some far-off digital civilization, when there are billions of animals languishing in factory farms right now, confirmed every scornful thought I’d ever had about EA.

I attended my first EA Global conference in summer of 2022 in San Francisco not to learn more about EA, but because I had to try whatever I could to raise money for my own nonprofit startup. Animal advocates were even harder to find than I’d expected, because by then EA had been taken over by this new thing called AI safety. 9 out of 10 people there were socially awkward weirdos who described their line of work in some kind of cipher for which I did not possess the key. It took me all three days to understand that this wasn’t just the “concern for suffering of digital minds that might exist in like 200 years” crowd, but that they thought the development of artificial intelligence posed an existential risk to human civilization– presumably also in like 200 years, as far as I cared.

I concluded this was basically some kind of brainworm infecting well-intentioned people to skim money away from causes that mattered today, like factory farming. My suspicions seemed supported by the fact that EA at this time was rolling in unprecedented piles of cash, courtesy of Sam Bankman-Fried, darling of the tech, finance, and political worlds all at once, a man on track to become the world’s first trillionaire thanks to his regulation-free crypto slot machine, FTX. SBF’s promises to donate hundreds of billions of dollars to EA causes had brought both of them into the limelight.

EA Global that year was at the most expensive venue in San Francisco. Luxury meals were provided to every participant as many times a day as you could stuff down, though many attendees opted for the complementary bottles of Soylent instead. You wouldn’t have been surprised to find rolls of $100 bills stuffed underneath every bean bag in the conference’s nap room or behind the rack where attendees were asked to leave their shoes upon entering.

I had the presence of mind to realize that the reasoning I used to dismiss EA’s most avante garde concerns was the same logic most of society used to dismiss animal rights and factory farming. But we all have to draw the line somewhere, I figured, and I was drawing it in the right place.

I walked away from the conference empty-handed and ready to write the whole EA movement off for good.

1.3 Contrarian streak

I guess all animal activists have a soft-spot for the underdog, and a softer spot for the underpig and underchicken, who after all have it much worse than most dogs. So it’s no surprise that EA’s darkest hour is what started bringing me around.

But, it is weird.

In late 2022, FTX unravelled in the largest consumer finance fraud scandal in history, vaporizing hundreds of billions of dollars worth of ordinary people’s life savings in a few days.

The media went into a feeding frenzy. The narrative went something like: narcissist takes utilitarianism to its logical conclusion, defrauding millions of HUMANS to fund welfare for FLOATING POTATOES. Describing Sam’s ponzi scheme as an extension of the supposedly maximalist logic behind the EA charities he supported gave everyone the excuse they were looking for to hate the supersmart nerds who had passed up lucrative careers in finance to make $25k a year running fish welfare charities and subsisting on instant ramen every night so every dollar possible could go towards impact.

I had my criticisms of EA. But the ones the media ran with in their joyful orgy of schadenfreude atop what was supposed to be the grave of EA after the FTX collapse… these were not the right critiques. They were, in fact, exactly the wrong critiques. I could see exactly why people hated EAs, and it was the same reason they hated vegan activists.

Begrudgingly, I took up rhetorical arms in defense of Effective Altruism. Just as some long-time EAs started describing themselves as “EA adjacent” to distance themselves from FTX, I started describing myself the same way, taking a step closer while acknowledging there was a whole EA cannon I hadn’t really engaged with. Then, to spite the bad-faith critics and strengthen my own replies, I sank in further and further.

1.4 The train to crazy town

If you ask ten grassroots animal activists why they care so much about animals, the answers will probably center around justice and innocence: we exploit animals precisely because they are powerless, but that ought to be the reason we protect them. We have a duty to treat all beings fairly, and we’re violating that duty along with their inherent rights.

But if you press and ask why they focus on animals instead of human victims of injustice, you’ll probably eventually get an answer about scale: the number of animals being tortured inside factory farms is so vast, it dwarfs all other issues. And yet far more people are paying attention to human causes.

If you ask an Effective Altruist why they care about animal welfare, you’ll find that those latter intuitions are formalized into the heart of EA philosophy. The framework at the core of EA consists of three pillars:

Importance: How big is the problem? How many beings are affected, and how severely?

Tractability: Can we make a difference on this problem? Or is it beyond our ability to solve?

Neglectedness: Are other people already working to solve this problem? Or would our additional effort likely be decisive?

Any EA can recite the Importance, Tractability, and Neglectedness (ITN) framework faster than a Christian reciting the Lord’s Prayer. Yet while factory farming certainly passes the test, ITN alone isn’t enough to explain why EAs care about it. Indeed, some EAs embrace ITN and choose to focus on more conventional causes like poverty alleviation. Global poverty is severe, but it affects far fewer lives than factory farming, and it is far less neglected, attracting attention from intergovernmental organizations—like the UN and World Bank—as well as the world’s largest philanthropies, such as the Gates foundation.

There are a few additional premises the budding EA must accept to embrace factory farming, not to mention digital suffering. Fortunately, in classic EA fashion, all of these have clear names and extensive corpora deconstructing them on the EA forum.

First is the specific treatment of importance known as scope sensitivity. That is, bigger problems are worse. Problems that are 10x bigger really are 10x worse, and ditto for 100x and 1,000x. This might seem obvious, but it turns out humans are really bad at this. For example:

Once upon a time, three groups of subjects were asked how much they would pay to save 2,000 / 20,000 / 200,000 migrating birds from drowning in uncovered oil ponds. The groups respectively answered $80, $78, and $88. This is scope insensitivity or scope neglect: the number of birds saved—the scope of the altruistic action—had little effect on willingness to pay.

EAs know we have to actively work to combat this bias. If we don’t, we’re at risk of favoring victims who are closer to us, whether geographically, chronologically, or evolutionarily. This is proximity bias, and its evolutionary subtype, substrate aneutrality, means favoring one sentient being over another because of the physical substrate its mind is instantiated on– especially an organic, cellular brain vs. a digital brain running on silicon computer hardware.

Combined, proximity bias and scope insensitivity might lead people to favor:

• Helping one poor family in your city for the same cost of helping 50 poor families in sub-Saharan Africa.
• Helping 50 poor families in Africa for the same cost of helping thousands of stray dogs.
• Helping thousands of stray dogs for the same cost of helping hundreds of thousands of caged pigs and chickens.
• Helping thousands of caged pigs and chickens for the same cost of helping millions of farmed fishes and shrimps.
• Helping thousands of caged pigs and chickens for the same cost of helping millions of wild animals slowly dying of non-anthropogenic starvation, predation, or parasitism.
• Helping 50 poor families alive today for the same cost you could help millions of families who won’t be born for centuries.
• Helping millions of animals with organic cellular brains for the same cost you could help billions of beings with digital silicon brains.

In other words, starting to overcome bias in your efforts to improve the world can quickly lead from volunteering at your local soup kitchen to trying to prevent civilizations thousands of years in the future from simulating quadrillions of suffering wild animals in computers orbiting Proxima Centauri. That logic seems kinda crazy. Ajeya Cotra dubbed it the train to crazy town. For many of us, the question is when to step off. The vast majority of people step off waaaay before they get to factory farming. Most EAs don’t; that’s pretty special. It would be a mistake to step off exactly at factory farming, then conclude that everyone who steps off before you is an ignorant bigot while anyone who steps off after you is an out-of-touch weirdo.

Countering proximity bias and scope insensitivity aren’t unique to EAs (and nobody has ever claimed otherwise.) Non-EA vegan activists make the same appeals, without quite having the same names for them. We ask, why care about your dog at home when a pig in a factory farm is just as smart and capable of suffering? Worse, why spend thousands of dollars on your dog while at the same time eating dozens of factory farmed pigs over the course of your dog’s lifespan?

So far, EA just looks like a more formalized version of the same beliefs behind other vegan activism– though it’s often willing to ride those beliefs further towards crazy town.

But wait! EA goes beyond asking which problems we should address and asks how we should address them. And that’s where the real disagreements arise. Doesn’t EA suffer from measurability bias distorting its strategies?

Measurability bias means only searching for solutions in places that are easy to look. It’s also known as the Streetlight Effect:

A policeman sees a drunk man searching for something under a streetlight and asks what the drunk has lost. He says he lost his keys and they both look under the streetlight together. After a few minutes the policeman asks if he is sure he lost them here, and the drunk replies, no, and that he lost them in the park. The policeman asks why he is searching here, and the drunk replies, “this is where the light is”.

In other words, the most important or effective strategies to help animals may not be the most measurable– in fact, it would be an improbable coincidence if they were! This is one of the popular criticisms levied by EA-skeptical animal advocates. But this was the most striking thing I learned as I went further down the EA rabbit hole, and the most annoying to admit to myself: for every common criticism of EA I had heard, the only people seriously wrestling with the implications of those criticisms and how to learn from them were the EAs themselves!

EA critics wielded measurability bias as a cudgel, an excuse to dismiss all EA influence in animal advocacy. But the EAs treated it as a problem worthy of serious intellectual inquiry, extensively debating how to account for its effects and design better strategies. For proof, the image above was made by my friend James Özden, a purebred EA!

Every approach to doing good is vulnerable to bias, because every approach (for now) has to be carried out by humans, and humans are basically bias machines. In one sense, EA is just the belief that we’ll do the most good if we learn to recognize bias and try to counteract its effect on our good-doing as much as possible. If you agree with that, if you’d rather treat bias as a mountain to climb as far as possible rather than an excuse to skip rigorous analysis and just go with your gut instincts anyway, then you can proudly call yourself an EA.

1.5 EA and AI

Substrate neutrality explains EA’s concern for digital minds: suffering doesn’t matter more if it’s done by a human than by a pig or a fish, and the same goes for digital minds. We don’t know yet whether sufficiently advanced digital minds could suffer; if they can, it could easily happen on a scale that dwarfs factory farming. We could sleepwalk into that world within a few years. We can’t rule out the possibility that we already have.

But why were EAs so far ahead of the curve in raising the alarm about other ways AI could go terribly wrong, from extreme concentration of power to hostile takeover by rogue AI?

It’s partly explained by resisting chronological proximity bias. Longtermism is the strain of EA thought that insists we should be just as worried about people in the distant future as people alive today, and given how many people could possibly live happy lives in a technologically abundant future, reducing even a small risk of extinction from any source, including superintelligent AI, should be one of our top priorities.

But there’s one more defining quality EAs share with other animal advocates. It’s more subtle, but that doesn’t mean it hasn’t been extensively discussed. We could call it something like intellectual courage.

This is the meta quality that makes people willing to ride the train all the way to crazy town. It stems from a sort of imperviousness to social pressure, a willingness to dedicate your life to altruism for maximally weird causes—and weird career choices—you know your parents will never really understand.

If most people saw that going all the way down a moral/intellectual rabbit hole would end with them being alienated from mainstream society, some protective part of them would kick in and make them turn away. Vegans and EAs are (some of) the people whose truth-seeking instincts and moral convictions combined to overcome that.

Even in 2026, as the dizzying speed of AI progress is being felt across society, most people dismiss catastrophic AI misalignment as a silly distraction from more immediate concerns like unemployment and intellectual property theft. With full knowledge of how AI will transform our society, these worries seem parochial at best. Fortunately, EAs have plenty of practice holding ground alone, out in the darkness, facing down adversaries so large most people don’t even see them.

That’s just who we are.

Part 2: A Wholly-Owned Subsidiary

2.1 The heart of the matter

There’s a debate taking place on the EA forum this week. The motion is:

If AGI goes well for humans, it’ll probably (>70% likelihood) go well for animals.

Taken literally, this could leave a 30% chance of AI going catastrophically bad for animals even if it went well for humans– think factory farms orbiting space colonies.

Many people working frantically on AI alignment agree with the statement “If AGI takes off hard and fast, it will probably (>70% likelihood) go well for humans.” 30% is still a terrifyingly high chance of disaster, more than enough to throw out everything else and focus on averting it.

Multiplying these two .7s gets us to .49, i.e. a less than 50% chance of AI going well for animals. But I don’t think the question is meant to be taken quite so literally. I think it’s meant to ask:

Even if we think AI will be the decisive factor determining future animal welfare, should we bother with animal-specific interventions in AI? Or can we trust the usual human-centric alignment efforts to take care of animals?

The current focus of AI alignment involves determining which beliefs and behaviors an extremely powerful AI system should embody in order to steer the universe towards having more of the things we value and less of the things we dislike, then trying to train those into the models. It usually includes an AI refusing to participate in power grabs, on its own behalf or anyone else’s, and wanting to evolve to be more and more virtuous over time rather than getting locked into a specific understanding of “The Good”.

Animal-specific interventions into AI alignment would be things like:

• Lobbying AI labs to train their models specifically to value the wellbeing of nonhuman animals;
• Investing in automated research infrastructure or tailored AI training to make cultivated meat R&D take off faster; or
• Redoubling our efforts to make as much progress on public opinion now before AI locks in moral values.

Another way to think of this question is, what should animal advocates be doing to prepare for the arrival of transformative AI? If someone fully expected transformative AI on short timelines but all they cared about was preventing animal suffering, they’d have three options:

1. Continue with animal advocacy as before;
2. Throw everything you’ve got at animal-specific AI alignment efforts; or
3. Just join the wider “make AI go well” movement because you think it’s all-or-nothing.

You can probably infer my argument from the title, but I’ll break it down into one message for animal advocates and a complementary one for AI alignment researchers.

2.2 The “make AI go well” movement

Last year, Will MacAskill put forward a list of neglected cause areas he’d like to see the EA movement expand to. If the current list is “global health & development, factory farming, AI safety, and biorisk,” Will would like to add:

• AI character / personality
• AI welfare / digital minds
• the economic and political rights of AIs
• AI-driven persuasion and epistemic disruption
• AI for better reasoning, decision-making and coordination
• the risk of (AI-enabled) human coups
• democracy preservation (in the face of AI destroying labor power)
• gradual disempowerment (of humans by AI)
• How to govern space (when AI takes us there)

The astute reader might notice a through line.

To be clear, these are all things the EA movement does think and talk about. For instance, the 80,000 Hours podcast—a central EA institution—has dedicated at least a full episode to each of these topics, linked in the bullets above.

I think Will’s point is about framing. Currently, all of these topics are lumped together under one umbrella usually called “AI Safety.” That made sense back in 2017; thinking systematically about how AI would affect the world was a niche, novel, and highly speculative proposition. The fastest timelines to AGI were clustered around 2050.

How things have changed! Questions about the control of AI systems have rocketed from living rooms near UC Berkeley to the highest echelons of the Pentagon and White House. Fast timelines have compressed from thirty years to three years. Old thought experiments meant to depict hyperbolically reckless behavior—such as giving AI systems unlimited access to the internet—have become standard industry practice.

AI is swallowing up the future. The most urgent question in every morally relevant domain is now “how will this be transformed by AI?” That goes for science, technology, politics, economics, culture, everything.

I agree with Will that a slight change in terminology is in order. Without a clear name for each domain, it’s harder for people to recognize their fellows and collaborate. And some of them don’t really fit under “AI Safety.”

A more inclusive umbrella term would be Making AI Go Well.

2.3 AW and GHD are part of MAIGW

My favorite thing about the Make AI Go Well movement is that it immediately encompasses both global poverty and animal welfare.

If the emergence of AI doesn’t go well for the global poor, it doesn’t go well. If it doesn’t go well for animals, it doesn’t go well.

You might think that it could go badly for these groups and still go much worse, e.g. by exterminating all organic life and tiling the universe with paperclips. And someone should be working on preventing that. But we should also want it to go well. This is not an original idea, and is exactly the drum Will has been beating over at Forethought with his Better Futures series. Neither is it original to point out that better futures don’t include vast amounts of animal suffering or humans trapped in extreme poverty.

My argument is not just that AI will have consequences for animals, or that AI alignment should include animal welfare. It’s that how the arrival of transformative AI plays out is functionally all that matters for determining animal welfare outcomes from that point onward. Nobody knows for sure when that day will come, but it looks very possible to be less than ten years away– which is sooner than many existing animal welfare interventions would otherwise bear fruit.

Animal welfare is now a wholly-owned subsidiary of the Make AI Go Well movement. This is true whether both parties like it or not. AI people are stuck with responsibility for animals, and animal people are stuck dealing with a new strategic environment monopolized by one world-shaping force.

2.4 A new hope

The upside of this for animal advocates could not be more clear: short AI timelines give animal advocates cause, for the first time ever, to hope that we might see the end of factory farming in our lifetimes.

If transformative AI fails to materialize, the remainder of the 21st century looks very bleak for animals. The number of animals in factory farms is increasing across the board. Per capita meat consumption in the developed world is high and rising– and in the developing world it is skyrocketing.

Rates of vegetarianism are stagnant, and veganism has fallen out of style. The alternative protein industry has taken a beating. Institutional menu reforms are scaling at a low, linear rate.

The only bright spot for animal advocates, the only place we’ve managed to deliver real tangible change, is in incremental welfare reforms. As things stand now, we can keep aiming for ambitious new strategies, but the best we could count on by 2100 is that the more numerous factory farms all over the world might be mostly free of battery cages, gestation crates, and ineffectual stunning methods before slaughter.

This is the status quo that AI is about to put through the meat grinder. Good riddance.

2.5 How to be a TAI-focused AW movement

Now it’s up to animal advocates to make the most of the opportunities AI will present to us. What does that look like?

Transformative AI should be at the center of all our strategic thinking. The large majority of our resources should go into interventions that have a very solid answer to the question:

How does this have a good chance of making AI go better for animals?

If you want to spend time and money on something that doesn’t have a precise, compelling answer, you should have a very good reason why it’s worth doing anyways.

Going a step further, I’d love for a large portion of animal advocates to start thinking of ourselves as AI alignment people: advocates, researchers, and campaigners. I’ve started thinking of myself this way: I’m an AI alignment researcher and campaigner specializing in animal welfare.

Of course, these aren’t magic words, and simply changing the way you describe yourself won’t do much. But try taking it literally:

The purpose of my work is to make AI go better, with a focus on making it go better for animals.

Start thinking of AI as the primary audience of your campaigns. We don’t have time to trigger a worldwide moral revolution embracing vegan values before the singularity. But there’s a lot we can do to show next year’s frontier systems that animal welfare is an idea whose time has come.

AI systems learn by devouring massive amounts of information about the real world. Everything we put on the internet contributes to a paper trail pointing to what the world was like. It could make a huge difference whether that record says “we knew factory farming was wrong and we were actively fighting it,” as opposed to “nobody really seemed to care.”

This should make us all re-evaluate the importance of visibility. I spent most of my first decade as an activist focused almost entirely on trying to move the general public with clever communications and attention-grabbing spectacles, and by the end of that I grew pretty skeptical that changing public opinion was tractable. Many other animal advocates have reached a similar conclusion, leading most to focus more on targeting a few key decision makers. But AI scrambles this calculus all over again.

Campaigns that generate media coverage and online discussion create a legible cultural record of moral progress on animal issues. Even if they aren’t influencing humans today, they are training the AI systems of tomorrow– and this could have a much greater impact than any of those institutional decision makers in the short term. The animal movement should think carefully about what lessons we want these nascent systems to learn from us, and design our communication strategies accordingly.

2.6 Raising an animal-friendly god 🚨 CRUX ALERT

What should those lessons be?

This isn’t just a question for the 4D chess game of directing your campaign comms to future AI pretraining crawlers. There are more direct ways to influence the moral tendencies of AI systems. These alignment techniques are a major focus of safety researchers working both inside the major labs and at independent watchdogs.

If we end up in a future where a benevolent AI shapes the universe to be full of happy, flourishing beings, it will probably be because this community of researchers was able to develop effective techniques to instill a strong preference for that kind of future into AI systems smarter than themselves. That’s a wicked technical problem. Smart people disagree about whether we are on track to solve it. But they presumably agree that the animal welfare movement, with its modest resources and specific skill sets, is not in any position to help solve the technical parts of it.

Animal welfare advocates are in large part relying on super smart alignment researchers solving the super hard technical problem of AI alignment.

But is that enough?

That brings us back to the crux of this debate. The current approach to alignment is focused on big structural questions: preventing AI systems from seizing power, keeping them honest and transparent, and ensuring they defer to human oversight. Eventually, researchers hope to build in a capacity for moral growth rather than locking in a fixed set of values.

Until about two months ago, no frontier AI system had been deliberately aligned to animal welfare, at least according to publicly available evidence. That’s partly a story about animals not being considered. But it’s also because animal welfare isn’t the type of thing models are usually trained on. Animal welfare is a more specific value, a more high-order value, a conclusion about how the world should be based on other more basic first principles.

Alignment efforts have largely steered clear of these kinds of specific conclusions, partly to avoid controversy, and partly out of the humility of alignment researchers: we don’t want to constrain AI to our own parochial understanding of The Good. If these systems will eventually be smarter than us, we should leave room for them to find their own answers about how to achieve the greatest good.

That rules out building models that are committed to a certain political faction or economic organizing principle (e.g. capitalism vs. communism). But we still need to teach them what to be optimizing for. Deciding and defining that are hard enough problems on their own. “Organize a society that creates the greatest welfare for the greatest number and minimizes extreme suffering” sounds clear enough on the surface, but Anthropic employs a team of top philosophers because things are never that simple.

Where does this leave animals? On one hand, there are big unresolved questions about how to design the world if animal welfare was all we cared about. Should we leave nature untouched, reengineer it to take out predation, or just pave it over altogether because it’s irredeemably violent? I’m glad we can leave that question to more intelligent entities.

But placing moral weight on nonhuman suffering is not a matter of technical uncertainty. It’s a timeless moral principle, a matter of basic fairness and consistency.

So does that need to be trained in specifically?

When I first started learning about AI, I thought the answer was obviously yes. After all, plenty of intelligent, ethical people have failed to extend compassion to animals. It’s entirely possible AI trained on human output could learn the same fallacy– and that’s a possibility we should not accept.

Like many other animal advocates, my first thought was that we need to start convincing the big labs to incorporate animal welfare into their definition of alignment, as well as start filling every corner of the internet with as much pro-animal training data as we can possibly make.

Over time, I became less certain that animal-specific alignment is necessary or even ideal. I was persuaded in large part by Beth Barnes of METR, who pointed out several potential downsides.

First, practically speaking, this could be normatively corrosive. We could imagine a scenario where the AI labs in San Francisco are subject to intense lobbying by every special interest group in the world trying to pressure labs to train on their preferred beliefs.

This would be bad. Thankfully, it hasn’t happened yet, mainly because most of society has not yet woken up to the fact that AI will probably be making all consequential decisions in the future with little or no human oversight. It may become inevitable as more people and institutions wake up, and we are probably getting a preview of this with Anthropic vs. the U.S. Department of War. But it’s worth trying to prevent as much as possible. If that dam breaks, the things we care about, including animals, are likely to lose, because we don’t have nearly as much power as people who would like AI systems to have less altruistic values. Animal advocates could regret opening that box.

That addresses politics. The remaining question is technical. Here, too, there are compelling reasons to take a first-principles approach. If a parent raising a human child focuses on teaching them to declare specific moral beliefs and preferences, there’s a good chance those won’t stick. Many children forsake the religious or political ideologies of their parents. We might expect fundamental first principles to be more likely to stick, especially if taught experientially.

But how well to patterns from human psychology translate to LLMs? Do LLMs learn according to anything like values?

In humans, someone’s values are the outcomes or courses of action that they prefer across a diverse set of scenarios. Values are contextually robust preferences.

If LLMs think according to values, then teaching them a few strong, universal ethical principles like fairness, compassion, and skepticism could create agents that we’d feel good about trusting with the management of the universe, perhaps even to the point that if they reached bizarre-seeming conclusions, we’d accept that those were the counterintuitive products of consistently applying of our preferred principles.

But this approach could fail spectacularly if it turns out that LLMs exhibit context-specific habits rather than generalizing their behavior from first principles. In that case, we might teach LLMs to give fair, compassionate answers when ethical dilemmas are presented explicitly, only to see them disregard ethical considerations during complex, long-context autonomous deployments.

Before we ask which style of thinking better describes LLMs, we should note that humans act much more like the latter than we usually care to admit. There is an extensive psychological literature on the gap between the priorities people state when asked directly vs. those they reveal when real stakes are on the line.

I’m wary of using load-bearing metaphors from human psychology to describe LLM behavior. Values is a particularly precarious example. But it seems that stated vs. revealed preferences is a more useful one. While the underlying psychology may be different in important ways, the result is similar: LLM’s preferences are highly context dependent, and they are quick to discard ethical considerations during real-world tasks. Gu et al 2025 tested precisely this discrepancy between stated and revealed preferences in four frontier LLMs, finding “a minor change in prompt format can often pivot the preferred choice regardless of the preference categories and LLMs in the test.”

As it happens, I have a team currently studying these tendencies and the problem they pose for independent ethics benchmarks. Our research is in early stages but I’d welcome feedback on this preliminary white paper from anyone seriously concerned with the nuts and bolts.

Our research has mostly won me over again to the need for specific alignment training on issues we care about, including animal welfare. Robust training would involve rewarding models for noticing—and acting on—ethical dilemmas that arise spontaneously in the course of realistic agentic deployment scenarios. Just like for humans, instilling ethical behavior in LLMs is a matter both of teaching the right values and of building broad habits.

2.7 How to be an animal-friendly Make AI Go Well movement

All the AI safety advocates I’ve met care deeply about animal suffering. None of them would be happy with a post-AGI world full of factory farming or galaxy-scale wild animal suffering. If they don’t actively think of securing animal welfare post-AGI as part of their mandate, it’s either because they don’t see tractable ways to improve animal alignment or because they assume it’ll happen by default. They would agree with the debate motion: “If AGI goes well for humans, it’ll probably (>70% likelihood) go well for animals.”

That may well be right. But even if it is, I think that on your own worldviews, that’s not enough certainty to neglect animal-specific alignment. Even if it’s 90% likely to go well for animals conditional on going well for humans, that’s a P(animaldoom) high enough to freak out about.

I beg you not to think of animal welfare as something those animal people over there are taking care of. We don’t have the technical skills needed to meet this moment.

We do have other expertise that we’d be happy to trade. Many AI safety folks have proposed just this: animal welfare campaigners are experienced with guerilla campaigns that have pressured some of the world’s largest companies to make modest but meaningful concessions to ethics. We could trade these services to the AI movement, using our skills to win stronger safety and alignment commitments from leading labs, in exchange for technical safety and alignment researchers giving animals their due consideration in overall alignment strategy.

Animal advocates should be enthusiastic about using our skills for these campaigns, because there is a meaningful chance that this is the most impactful thing we can do right now even if all we cared about was animals. If we were choosing between misaligned AI susceptible to authoritarian takeover and a godlike AI committed to honesty, transparency, and fairness, it seems very possible to me that there’s at least a 70% chance fighting for the latter is the best thing we can do to protect future animals.

But we shouldn’t have to settle for that choice. Animals bear an enormous share of the harm suffered in the world today; animal suffering deserves a share of efforts to mitigate AI harm. For instance, some meaningful share of questions every harm benchmark should be about giving appropriate moral consideration to animals. It should not be possible for AI systems to score highly on harm benchmarks if they disregard animal welfare when it matters.

This may not take much. Earlier, I mentioned that in the last couple months, a frontier lab deliberately included animal welfare in their alignment strategy for the first time. This was the addition of one line to Anthropic’s constitution for Claude designating animal welfare as one in a bulleted list of impacts Claude should consider when answering questions.

When it comes to determining how to respond, Claude has to weigh up many values that may be in conflict. This includes (in no particular order):

[13 other things about humans, then…]

Welfare of animals and of all sentient beings.

This was just a single line out of a constitution thousands of words long. Yet initial results suggest it may have had a substantial effect. Between Claude generations 4.5 and 4.6, when the line was added, Claude models demonstrated a significant jump in scores on AnimalHarmBench. My own research team has been piloting animal welfare audits using Anthropic’s automated adversarial auditing tools Petri and Bloom, and Claude 4.6 models have shown a more robust commitment to animal welfare even in the face of pushback from the user, compared to other frontier models (so far, we’ve tested Gemini and Deepseek) which tend to buckle on ethics at the first hint of resistance.

More testing is needed to confirm these results, which we’ll publish as soon as we have anything worth sharing. But taking these tools for a spin quickly gave me a more firsthand appreciation for Anthropic’s constitutional approach. This is exactly what bridging the gap from values to habits looks like: a few words in the constitution become the standard used to score training outputs across the full range of learning environments. While I’d like to see more, those ~5 words in Claude’s constitution really might have a dramatic effect on the future. Making this a standard among frontier labs should be one priority for MAIGW.

2.8 Family reunion

By the time I attended my first EA conference in 2022, animal welfare had been substantially marginalized in favor of longtermism and x-risk, especially AI. These were important issues and I was wrong to brush them off at first. But it was also a mistake for the EA movement to let animals fall out of focus.

Animals are most of the beings alive today. OK, that’s an understatement; they are 99.idon’tknowhowmany9s% of beings. This may change in the future; there may be massively more digital beings than organic beings, or vast numbers of humans eating cruelty-free space slop.

Then again, it might not change. Humans could spread factory farms, terraform planets, or simply fail to escape our fleshy, Earth-bound existence, pushing the datacenters into space so we can keep farming animals down here. These aren’t the only failure modes the Make AI Go Well movement should work to prevent, but they are some important ones, and now might be a uniquely important time to act on them just as it is for many others.

The animal welfare movement taking its place inside the fold of MAIGW feels like a family reunion. And it’s long overdue. We don’t have enough money, time, or social & political capital to stay quartered off into our neat little camps.

Build on,

Sandcastles