A Dog, an AI Chatbot, and a Custom Cancer Vaccine: What the Story Actually Shows

The clip you probably saw

A short video makes a simple claim: a man designed a vaccine that cured his dog's cancer. The real story is more interesting than the caption, and it ends differently.

The dog is Rosie, an eight-year-old rescue Staffordshire bull terrier cross in Sydney with aggressive mast cell cancer, a common skin cancer in dogs. Surgery and chemotherapy had not held it. Her vets estimated she had months.

Her owner, Paul Conyngham, works in tech and machine learning. He did not accept the estimate, and he did not make a vaccine in his kitchen. What he did was closer to project management across a set of tools and specialists.

What was actually done

The steps, as reported by UNSW and by news outlets covering the case:

• Sequencing. He paid a university genomics lab to read the DNA of Rosie's tumor and of her healthy tissue, and compare them. The difference is a list of mutations.
• Picking targets. He used AI chatbots and the protein-structure tool AlphaFold to sort that list, looking for mutations that would produce protein fragments unique to the tumor. Those fragments are what NCI calls neoantigens - new proteins that appear on cancer cells but not on normal cells.
• Building it. He took the shortlist to scientists at the University of New South Wales, including the UNSW RNA Institute. They reviewed the data, designed the mRNA, and manufactured it in a lab, wrapped in lipid nanoparticles - the same delivery approach used in mRNA COVID vaccines.
• Giving it. Rosie received the one-off vaccine at a veterinary research centre in December 2025, with booster doses after. She was also given an immune checkpoint inhibitor, a separate kind of immunotherapy.

Within weeks her owner and vets reported that several tumors had shrunk noticeably, and that her energy and mobility improved sharply. Reporting described her going from barely moving to chasing a rabbit.

What happened next

Not a cure. That word came from the people involved, not only from critics. Some tumors never responded, and a second round of analysis was started to target a different set of mutations.

In August 2026, Conyngham launched a startup, Gamgee, to build personalized mRNA vaccines for dogs, with a US$4 million seed round led by Founders Fund and clinical trials beginning in Australia. The company's own launch materials, as reported in August, said Rosie entered remission but that her cancer returned after a later surgery.

That ending matters, and it is missing from most of the viral versions of the story.

Why one dog cannot settle anything

This was a single animal treated outside a trial. Several things make the result impossible to interpret cleanly:

• No comparison group. Nothing was run alongside Rosie to show what would have happened without the vaccine.
• Two treatments at once. She received a checkpoint inhibitor as well. When two things are given together and something improves, neither one has been shown to be the cause.
• Mast cell tumors are erratic. They can wax and wane on their own, which is exactly the pattern that fools an audience of one.
• Key details were never published. Tumor grade and other clinical specifics have not been released, so outside experts cannot judge the case.

Writing in The Conversation, oncologist Justin Stebbing of Anglia Ruskin University made the point plainly: the AI did not do the curing. It worked as an always-available guide, and qualified scientists still checked its work and did the laboratory work.

Where the human science actually stands

Personalized neoantigen vaccines are real research, not science fiction - and they are still early. NCI reported in 2025 on two small trials: nine kidney cancer patients at Dana-Farber, and sixteen pancreatic cancer patients at Memorial Sloan Kettering. Immune responses lasted years in some patients and side effects were mild. NCI also noted the obvious limit: these were tiny studies, larger trials are underway to confirm the findings, and in kidney cancer long remission after surgery is common anyway, so credit is hard to assign.

For treatment vaccines that have finished that road, the list is short. NCI names sipuleucel-T for advanced prostate cancer and T-VEC for melanoma that has come back after surgery. Everything else remains investigational.

Our page on personalized cancer vaccines covers what they are and why they stay experimental, and our overview of immunotherapy explains the wider family of treatments this belongs to.

What this story is good for

It is a fair illustration of something genuinely new: sequencing, mRNA manufacturing, and AI assistance have each become cheap and accessible enough that a determined non-scientist can assemble them, with a university lab doing the parts that require one. That is a real shift, and it is the part worth paying attention to.

It is not evidence that any treatment works. Evidence comes from clinical trials with comparison groups, pre-set measures, and published results.

If a headline like this reaches someone you love

• Ask whether the report describes a trial or a single case.
• Ask how many patients were treated, and whether anyone was compared.
• Ask what else the patient was receiving at the same time.
• Ask whether the results have been published where other experts can check them.
• Bring the article to the treating team rather than acting on it. Our guide to biomarker testing explains the tumor testing that any personalized approach starts from.

How this article was prepared

An AI-assisted editorial system helped prepare this page. No named medical reviewer has reviewed it unless one is listed.

The National Cancer Information Foundation publishes Cancer Explained. This page is for learning. It is not medical advice and does not suggest a test or treatment. It is not veterinary advice.

See an error, old source, or unclear wording? Tell us.