One conceptual reality to consider is that the ants on the surface of any hill are not quite as sentient as the queen. The survival of a queen that produces viable progeny is paramount to the survival of the species and, at a more granular level, the bloodline of the individual colony.
The queen is dependent on the drones being as obedient as your fingers and tongue might be to your brain, with your brain being the queen organ, for your hundred(s of) pound(s of) tissue. We know that the surface ants support the colony, which supports the queen, and overall, the biomass of the ant colony is less than any single mammal, and we know that any signaling is provoked by scent, based on attack behaviors at a minimum.
Meaning that reaction to scent wafts through colonies quickly, engaging behavior as quickly as it is noticed. Ants seem to drop everything, and move, unanimously, when a signal is given. The signals must be pretty complex, with pointers hooked into fixed physical actions on some level. The ants probably don't think very much about what they should do next, as much as constantly listen or smell for a prevailing vote on what to do. Attack, forage, assemble into a formation, wait for further instructions, and so on.
In ant studies, an ant colony is segmented into chambers, each having odors, the most conspicuous being the colony's waste dump, so you have to figure that the colony survives, and expects the surface soldiers to go permanently missing once in a while, when scent signaling is disrupted.
These fungal organisms probably interact with a degree of equilibrium, when exposed to healthy ant colonies. Usually a parasitic relationship with a host is balanced to not completely destroy the host population, according to an almost intuitive self-preservation instinct among parasites. So the fungus probably only disrupts soldiers at the periphery of the colony, without destroying the core of the hill.
You can get a sense of this, based on the fact that the affected ants are driven away from the hill, toward environments more opportunistic to the fungus. If a version of the fungus were to grow opportunistically within the colony, it might invade and drive the ants to extinction, which might turn out to be, in effect, suicidal for the fungus too, and not just the ant.
The colony is really the organism, and the soldiers serve it, abiding by prevailing odors, likely without very much free will. Or at least as much free will as your arm has, to reflexively recoil from a burn, or lash out at some unassessed threat when reacting in fear.
This fungus probably is basically "numbing" the arm, but the wild part of seeing your arm suddenly behave with a mind of its own, would be an observation from the perspective of the queen, and is probably more or less like seeing a junkie or crackhead stumble around like a zombie, as far as the colony-as-organism is concerned.
So, if the ant is "fighting" the "will" of the fungus, it's probably more likely that the ant is receiving strong scent signals compelling it to act a certain way, and the sauce that the fungus is producing as an override to the biological signals in the ant's circulatory system (remembering that insect circulatory systems are primitive, and probably easily hackable fluid systems) might be weak, or the infectious load might be light, resulting in a noisy, conflicted signal at the neurological level. Maybe like a junkie, not completely ruined by dope.
tl;dr soldier ants follow their nose without a whole lot of agency, because they are born to serve the colony, selflessly. Thus they likely are unconcerned by a "loss of self," and perhaps might be more emotionally invested in failing the colony, if soldier ants have emotions.