AI-Powered Cancer Vaccine Shows Promise in Dog, Sparking Debate on Future of Personalized Medicine

A machine learning engineer in Sydney, Australia, has gained attention for developing a personalized mRNA cancer vaccine for his dog, Rosie, using artificial intelligence tools like ChatGPT and Google DeepMind’s AlphaFold. The experimental treatment led to a significant reduction in Rosie’s tumor size, raising hopes and prompting discussion about the potential of AI in veterinary and human medicine.

From Diagnosis to DIY Vaccine

Paul Conyngham adopted Rosie, a Staffordshire Bull Terrier-Shar Pei mix, in 2019. In 2024, she was diagnosed with mast cell tumors on her leg. Traditional treatments, including surgery, chemotherapy, and immunotherapy, provided limited success, and veterinarians estimated Rosie had only one to six months to live. 1 Determined to find a solution, Conyngham turned to AI.

He utilized ChatGPT to formulate a plan for tackling canine cancer at the molecular level, focusing on genomic sequencing, mutation identification, and vaccine design. 1 Conyngham then collaborated with researchers at UNSW’s Ramaciotti Centre for Genomics and the UNSW RNA Institute to sequence Rosie’s tumor DNA and produce the mRNA vaccine. 1 The genomic sequencing cost approximately $3,000 Australian dollars. 2

AI’s Role in Vaccine Design

Conyngham, with 17 years of experience in machine learning and data analysis, analyzed the genomic data and identified mutations of interest. 2 He used AlphaFold to predict the protein structures resulting from these mutations and matched them to potential drug targets. 1 The process culminated in an mRNA sequence formula, which was then used to create the physical vaccine. 1

Promising Results, Cautious Optimism

Following the first injection in December 2025, Rosie’s tumor shrank by 75% within a month. 1 Scientists at UNSW and the University of Queensland have confirmed the results, but emphasize that this is a single case with no controlled trials. 1

Expert Concerns and Future Implications

Researchers caution against overenthusiasm, highlighting the challenges of proving both safety and efficacy in clinical trials. Patrick Heizer, a researcher in cell and gene therapy, points out the difficulty in ensuring precision, as therapies targeting tumor proteins could also affect similar proteins in healthy organs. 1 results observed in animals may not directly translate to humans due to differences in protein structures. 1

Despite these caveats, the case demonstrates the potential of AI to empower individuals to engage in medical research, and development. 1 The story has garnered attention from AI leaders, including OpenAI President Greg Brockman and Deepmind CEO Demis Hassabis. 1

Conyngham began the project in November 2024. 1