He used ChatGPT to design a cancer vaccine for his dog — it worked

Paul Conyngham's dog Rosie was dying. The 8-year-old rescue had mast cell cancer — aggressive tumors spreading across her body. Chemotherapy wasn't working. Surgery couldn't keep up. Veterinarians said there wasn't much more they could do.

So Conyngham, a Sydney-based tech entrepreneur with zero medical training, did something unprecedented: he used AI to design a personalized cancer vaccine. Within weeks of the first injection, Rosie's largest tumor — roughly the size of a tennis ball — shrunk by 75%.

How One Man and Three AI Tools Designed a Vaccine

Conyngham started with ChatGPT. He asked it to help him create a plan — and the AI suggested he get Rosie's tumor DNA sequenced and pointed him to the UNSW Ramaciotti Centre for Genomics in Sydney. That sequencing cost $3,000.

Next came AlphaFold, Google DeepMind's protein structure prediction tool. Conyngham used it to identify which mutated proteins in Rosie's cancer could be targeted by a vaccine — essentially finding the weak spots in the tumor's armor.

Finally, Grok (Elon Musk's AI) designed the final vaccine construct — the specific blueprint for an mRNA molecule (the same technology behind COVID-19 vaccines) that would teach Rosie's immune system to attack the cancer.

The Scientists Who Made It Real

This wasn't a solo AI miracle. Professor Pall Thordarson, director of UNSW's RNA Institute and a pioneer in nanomedicine, took Conyngham's AI-generated data and turned it into an actual vaccine. Thordarson called it "the first time a personalized cancer vaccine has been designed for a dog."

The vaccine was administered alongside a checkpoint inhibitor (a drug that helps the immune system recognize cancer cells), making it impossible to know exactly how much the vaccine alone contributed.

Why Experts Are Cautious — and Why It Still Matters

The story went viral after OpenAI president Greg Brockman and DeepMind CEO Demis Hassabis both shared it. But researchers quickly flagged important caveats:

Patrick Heizer, a cell and gene therapy researcher, warned: "Fighting tumors in the lab is the easy part. The real challenge is proving a therapy is both safe and effective in controlled trials."

Egan Peltan, a Stanford PhD in chemical biology, estimated the true cost was $20,000–$50,000 and noted that conventional methods could have achieved similar results. The AlphaFold protein prediction had a confidence score of only 54.55% — described as low.

Conyngham himself is measured: "I'm under no illusion that this is a cure, but I do believe this treatment has bought Rosie significantly more time and quality of life." Some of Rosie's tumors didn't respond at all.

What This Really Shows About AI in Healthcare

Strip away the viral hype and there's still a genuinely important story here. A person with no biology background used free AI tools to:

• Navigate cutting-edge biomedical research in weeks, not years
• Identify specific cancer targets from raw genetic data
• Connect with the right scientists to build something real

Thordarson sees broader implications: "Personalized medicine can be very effective, and done in a time-sensitive manner, with mRNA technology." If AI can accelerate this process for a dog, it could eventually help humans — where personalized mRNA cancer vaccines are already in clinical trials at companies like Moderna and BioNTech.

As for Rosie? She's back at the dog park. Conyngham says she recently jumped a fence to chase a rabbit — something that would have been unthinkable months ago.

Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News