AI-designed cancer vaccines for dogs: what is real, and what exists today.

Did someone really use AI to design a cancer vaccine for their dog?

Yes, in a narrow and important sense. A widely covered case in Australia involved a dog owner using computational tools to help design a personalized mRNA cancer vaccine for his own dog, working with researchers and veterinarians. It was one dog, not a clinical trial.

That distinction matters more than anything else on this page. A single case, however moving and however carefully done, cannot tell you whether a treatment works. It cannot separate the treatment from the surgery that came before it, from the individual biology of that animal, or from chance. It is a proof that something can be built, not evidence that it works.

What the story did do is make a lot of people aware that personalized cancer vaccines for dogs are now technically possible. That is why you are probably here. (The case was covered by Reason, among others.)

What a personalized mRNA cancer vaccine actually is

It is a vaccine built from one specific tumor, in one specific animal. Rather than targeting something shared across all dogs with a given cancer, it is designed around the particular mutations found in that individual dog's tumor.

The underlying idea is not new or exotic. As a tumor grows, it accumulates mutations that healthy tissue does not have. Some of those mutations change proteins in ways the immune system could, in principle, recognize as foreign. Those changed fragments are called neoantigens. A personalized cancer vaccine is an attempt to point the immune system at them deliberately.

The mRNA part is the delivery method, and it is the same basic mechanism people became familiar with during the COVID vaccines: the mRNA carries instructions, cells read those instructions and briefly display the corresponding fragments, and the immune system learns to recognize them. The cells do not become cancerous; they display a marker, not the disease.

It is worth being clear that this is different from the routine vaccines your veterinarian already gives. Those are licensed products, given to healthy animals to prevent infectious disease. This is an experimental, individualized approach in an entirely different category.

What the "AI" actually does

It does the comparison and the ranking, not the medicine. No algorithm decides whether to treat a dog, and no software administers anything. The computational work happens well upstream of any clinical decision.

In practice it involves roughly three steps. First, sequencing the tumor and comparing it against the dog's own healthy DNA, so that mutations belonging to the cancer can be separated from the normal variation every animal carries. Second, using prediction algorithms to estimate which of those tumor-only changes the dog's particular immune system is most likely to be able to display and recognize. Third, assembling the highest-ranked candidates into a single designed mRNA sequence.

This is where personalization stops being a marketing word and becomes a technical necessity. In canine hemangiosarcoma, for example, published sequencing work has found that a substantial share of tumors, on the order of roughly half, carry no identified recurrent driver mutation at all. A strategy that only targeted a known list of common cancer genes would have nothing to aim at in those animals. Reading the individual tumor is the only way to find targets in a cancer that varies that much from dog to dog.

It is also where honesty is required about limits. These prediction algorithms were largely developed and validated on human immune biology, and canine immune genetics are less thoroughly mapped. Anyone working in this area, including us, is extrapolating to some degree, and that is a real scientific limitation rather than a footnote.

What is established, and what is still early

The human science is genuinely advanced; the canine application is genuinely early. Both halves of that sentence matter.

In human oncology, personalized neoantigen mRNA cancer vaccines have progressed into late-stage clinical trials run by major pharmaceutical developers, and the field is well past the question of whether such a vaccine can be manufactured for an individual patient. That is why the approach is credible as science rather than speculation.

In dogs, the picture is much earlier. There is no licensed personalized mRNA cancer vaccine for dogs. The published veterinary experience is limited, individual cases are not evidence of efficacy, and the honest state of knowledge is that nobody can currently tell you what survival benefit, if any, such a vaccine provides for a dog.

There are also real risks that deserve to be stated plainly rather than buried. Directing an immune response at a tumor carries a theoretical risk of the immune system also reacting against healthy tissue, and any experimental biologic carries unknowns that a licensed product does not. These are the kinds of risks that make sense to weigh only for aggressive cancers where standard options are genuinely poor, and only in consultation with a veterinary oncologist.

Can I get this for my dog in the US?

Not as a product you can buy today. There is no licensed, generally available personalized mRNA cancer vaccine for dogs in the United States, and any claim otherwise deserves real scrutiny.

What does exist is research-stage work. In our case, that means we operate as a computational and bioinformatics design service: we work from a dog's tumor and matched normal sequencing to design a candidate mRNA blueprint. We do not manufacture the product ourselves, we do not practice veterinary medicine, and we never administer anything. Anything of this kind would have to be given by the dog's own licensed veterinarian, under the appropriate USDA Center for Veterinary Biologics and state authorization, alongside and not instead of standard care.

We work case by case with veterinary oncologists, and we would rather tell an owner honestly that their dog is not a candidate than take on a case we cannot help. That has already happened, and it will happen again.

Who we are, and who we are not

We are an independent United States team, and we are not the group from the Australian story. We are not affiliated with that project, with the individuals involved in it, or with any company that has grown out of it. We were inspired by the same idea, and we have a great deal of respect for anyone who fights that hard for their dog, but we are a separate effort with our own team, our own methods, and our own responsibility for what we say.

We mention this explicitly because the confusion is common. Owners have reached out to us believing we were that team, and it is better to be clear than to let a hopeful assumption stand uncorrected.

What to do right now if your dog has cancer

The single most time-sensitive thing is usually the tumor tissue. If your dog is having or has recently had surgery, ask the surgical team and the pathology laboratory, in writing, to preserve the remaining tumor tissue and place a hold on it. Laboratories routinely discard specimens after a set period, and once tissue is gone, no sequencing-based approach is possible at all. We wrote a practical guide to preserving a tumor sample that covers what to ask for.

Beyond that: work with a veterinary oncologist, ask directly what standard of care realistically offers for your dog's specific cancer and stage, and ask whether any clinical trials are open to you. If you want to understand the disease itself, our explainer on hemangiosarcoma lays out what the published survival data actually shows.

And be careful with anyone who offers certainty. In aggressive canine cancers there is currently no cure, and the honest options are measured in added time and quality of life rather than in promises.