In a scientific article published a few days ago, Levin and colleagues show that aggregates of bacteria (Bacillus subtilis) are able to recognize whether frog embryos (Xenopus) or a Xenobot have been placed in the well together with them. The Xenobot is a small living construct made with embryonic cells of the same frog, but with a shape and behavior different from those of the natural embryo.
Left alone in the liquid, bacteria capable of moving form a branched network on their own. A frog embryo or a Xenobot (called “targets”) is placed in the well, and some of the bacteria gather around the target and form a bright halo, visible in the photographs. The target is moved: the halo re-forms in the place where the target has been put. It is removed: the accumulation dissolves. A live embryo makes the halo grow over time, as if more and more bacteria were interested in going around it. One killed by heat gives an initial accumulation that, however, dissolves shortly afterward, as if the bacteria gave it a first look but decided they were not interested because it is not alive. Likewise, minced frog tissue does not evoke a response from the bacteria. The bacteria remain bound to a living, intact object, and to its position.
Embryo and Xenobot both produce this halo. To the eye the two rings look alike. The finer difference, however, lies in how the bacteria occupy the rest of the well — more or less dense, more or less branched — a difference too subtle to be appreciated by the human eye. To measure it the researchers use an artificial-intelligence tool. The comparison is demanding because embryo and Xenobot share a genome and a cellular origin. What changes is their anatomy; the embryo has the normal anatomy of an embryo, the Xenobot is essentially a cluster of cells without anatomical organization.
In other words, the same living material, organized in a different way, is not the same object for the bacteria, which arrange themselves differently depending on whether there is an embryo or a Xenobot in the well. It is as if the spatial arrangement of the bacteria reflected the anatomy of the organism present in the well, embryo or Xenobot. And what is even more interesting is that this information acts at a distance. “At a distance” means that the bacteria do not have to infiltrate the tissue of the embryo or the Xenobot as in an infection: they reorganize in the liquid around it, through the shared environment, even far from the organism present in the well.
Levin’s group calls this inter-kingdom morphogenesis: a living collective writes part of its own state into the form of another. The bacteria function as living sensors: their collective organization reports the state of another being and distinguishes its nature.
Now consider a fact that makes the result even more instructive. The bacteria in the study were all of the same strain. There was no biodiversity: a single species, a genetically coherent population. And yet that poor collective, in the liquid of a well, distinguished at a distance an embryo from a Xenobot — two bodies made of the same material, but organized in a different way. If already a single strain can register the nature of what is near it, what may happen when the microbial community is not poor, but rich?
That is the question that arises for the Bravo community once it is ingested. There is not a single Bacillus there. There are bacteria, eukaryotes, phages, and plasmids: different kingdoms, different genetic languages, different ways of occupying the same environment. The “host” is not a frog embryo in a well. It is the human body, with its anatomy, its physiology, its variations from person to person, and sometimes its pathologies.
The analogy does not prove that Bravo “recognizes” a disease the way a physician makes a diagnosis. It does say, however, that the channel already exists in its simplest form: a microbial collective can reorganize itself according to the state of another living being through the shared environment. If this happens with a single strain and a tiny target, it is fair to ask how much more articulated the interaction between a multi-kingdom community and a human host may be. Perhaps that community does not respond to everyone in the same way. Perhaps it registers different needs — an inflamed intestine is not a quiet intestine, one metabolism is not another — and it adapts.
The manifold results that are reported every day, different from person to person, do not by themselves prove this reading. They do, however, make it thinkable. Where Levin sees sensor-bacteria around an embryo, one can imagine a sensor-community around a body: not a drug the same for everyone, but a living collective that, once received, enters into dialogue with the host that harbors it.
https://www.biorxiv.org/content/10.64898/2026.09.02.748853v1
