Epistemic status: Not particularly certain. I am very likely wrong about some or all of these points, and would like to hear counterarguments to these in the comments.

Many utilitarians and effective altruists advocate moral circle expansion as a cause area in order to reduce far future suffering. While I am a negative-leaning utilitarian, I am skeptical that moral circle expansion will result in less suffering in the future, and I could realistically see it contributing to active harm. Herein, I will discuss the reasons for my skepticism.
 

Orthogonality of values

For the purposes of this article, I will define a large moral circle as a utility function that extends to a large number of individuals, i.e. people with a large moral circle care about what happens to a large number of beings. The main problem I see with this is that people with a large moral circle are not necessarily utilitarians. Moral circle expansion combined with negative utilitarianism may be a good thing, but it will not necessarily be combined with negative utilitarianism in practice. A malevolent sadist who wants to cause as much suffering as possible could be considered to have a large moral circle, but their moral circle will result in making the world far worse if their values are put into practice. Religious fundamentalists who want to impose their morality on everyone also arguably have a large moral circle, but their values they want to force on everyone are based on religious principles rather than utilitarian considerations. 
 

The dominance of life-focused values

In order to determine the effects of moral circle expansion, we need to determine which set of values individuals with large moral circles will have in the long run. I would argue that the most adaptive set of values are values that assert that life is sacred, and which value preserving and expanding life as much as possible. In the long run, the effects of this may become far more pronounced.  The reason for this is simply because of the effects of natural selection. Individuals who value life will reproduce more than individuals who don’t. Religions like Christianity and Islam, for example, encourage their followers to proselityze their religions and promote natalist values, and both consider suicide to be sinful. Antinatalists generally do not reproduce, so any predisposition to antinatalism will tend to be weeded out of the population over time. Andrés Gómez Emilsson has argued that the future could potentially be dominated by “pure replicators'' that consist of valueless patterns that value replicating themselves and nothing else. Maladaptive values may only be able to exist within the Hansonian dreamtime, which will be supplanted by more adaptive value systems as the dreamtime subsides. 

A recent example of the effects of a large moral circle without utilitarianism is the overturning of Roe v. Wade. Anti-abortion activists and the Supreme Court justices who overturned it have a large moral circle as defined above, since they want to impose their values on an entire country. But they are not utilitarians with suffering-focused values, and are instead motivated by religious principles and the belief that life is sacred.

People with large moral circles will likely want to create large populations of sentient beings. From the perspective of an antinatalist or EFIList, creating large amounts of sentient life would be doing active harm by creating large amounts of harm. Many environmentalists want to increase wild animal populations and prevent species from going extinct, yet it could be argued that utilitarians should advocate wildlife antinatalism in order to decrease suffering in nature. Antinatalists may therefore want to promote moral circle narrowing rather than expansion.

Are large moral circles sustainable?

Another potential problem with moral circle expansion is that there is a risk that large moral circles are unsustainable. There is a significant risk that groups with large moral circles could collapse in the long run, and could be out-competed in the long run by groups with narrower moral circles. Universalism is a fairly recent development in human civilization, and throughout history, ethnocentrism has been the norm throughout most cultures. One study tried to determine the effects of different value systems as evolutionary strategies, including ethnocentrism, selfishness, universalism, and traitoriousness. They created a simulation to determine the effects of competing value systems, and which values would win out in the long run. In the simulation they created, they found that ethnocentric values tended to win out in the long run.

Moral circle expansion could cause psychological suffering

A final reason why I’m skeptical of moral circle expansion as a cause area is that it could result in significant psychological suffering for its adherents. This may be less relevant to longtermism than the other points discussed above, but is a relevant short term consideration. If people with large moral circles are high in affective empathy, any deviation from their values will likely cause them significant psychological suffering. Consider mass shootings, for example. Any disaster that draws national attention will cause mourning on a national scale. From a felicific calculus standpoint, the vast majority of the suffering generated is the result of numerous people feeling bad for the victims, and the suffering felt directly by the victims is only a small proportion of the total suffering generated. 

Consider a hypothetical society where everyone in the society is a clinical psychopath or sociopath. Ironically, it is likely that this society would have less suffering per capita compared to a society of empaths. Even if a hypothetical society of sociopaths would have more malevolence and criminality on average, such a society would also have a complete absence of affective empathy, guilt, and grief. Because of this, such a society may have far less suffering on average compared to a society higher in empathy.