TL;DR
- Debunked Claim: A viral story claiming ChatGPT cured a dog’s cancer misrepresented AI’s role, which was limited to research assistance.
- What Actually Happened: University scientists developed a personalized mRNA vaccine for the dog, requiring months of expert labor and $3,000 in genomic sequencing.
- Unproven Results: The vaccine was administered alongside a checkpoint inhibitor, making it impossible to determine which treatment contributed to any improvement.
- Hype Amplification: OpenAI president Greg Brockman and Elon Musk shared the story on social media without noting that the treatment remains unverified.
OpenAI’s ChatGPT did not cure a dog’s cancer. A viral story endorsed by OpenAI president Greg Brockman and Elon Musk credits AI for work that human scientists performed. But as The Verge reports, the treatment’s effectiveness remains unproven.
Newsweek, the New York Post, and several tech executives amplified claims that ChatGPT cured a dog named Rosie. In reality, AI served only as a research assistant while university scientists with thousands of dollars in resources did the actual work. Moreover, because the vaccine was administered alongside a checkpoint inhibitor, determining which treatment contributed to any improvement is impossible.
Rosie’s Treatment Journey
Paul Conyngham, a Sydney-based tech entrepreneur with no background in biology or medicine, learned in 2024 that his dog Rosie, a Staffordshire bull terrier-shar pei mix, had cancer. Chemotherapy failed to shrink the tumors, and vets told him nothing more could be done.
With conventional options exhausted, Conyngham turned to ChatGPT, which surfaced immunotherapy as a potential option and pointed him toward experts at the University of New South Wales. From there, Professor Pall Thordarson, director of UNSW’s RNA Institute, developed a personalized mRNA vaccine tailored to Rosie’s tumor mutations in less than two months.
Meet Rosie, my 7-year-old staffie x shar pei. Back in 2022, I noticed strange lumps on her head. What the vet deemed as “just warts,” ended up being late-stage cancer. They said nothing could be done but I refused to give up. So I took it upon myself to find a cure. Enter: AI. 🧠
– Paul S. Conyngham (@paul_conyngham) November 22, 2024
Subsequently, Rachel Allavena, a canine immunotherapy professor at the University of Queensland, administered the treatment at her Gatton laboratory. UNSW professor Thordarson told The Australian he believes it is the first time a personalized mRNA vaccine has been designed for a dog.
Despite the narrative of a chatbot-powered breakthrough, the process was neither quick nor cheap. Conyngham paid $3,000 for genomic sequencing and spent three months writing a 100-page ethics approval document. In an interview, Conyngham acknowledged the treatment is not a cure but said he believes it has extended Rosie’s life and improved her quality of life.
Furthermore, navigating Australian ethics approval for a drug trial on his dog, he said, proved more challenging than the vaccine development itself.
Rosie received the vaccine in December 2024 alongside a checkpoint inhibitor. A few weeks later, some tumors had shrunk but not disappeared entirely, and one tumor did not respond at all. With both treatments administered simultaneously, isolating the vaccine’s contribution remains impossible.
As a result, the cost, institutional access, and expert labor required to move from a ChatGPT suggestion to an injectable vaccine places this case far outside the reach of ordinary pet owners.
AI’s Role Was Minimal
Even setting aside the question of accessibility, the technical claims about AI’s contribution do not hold up to scrutiny. ChatGPT did not design or create Rosie’s treatment: Human researchers performed the actual scientific work.
Instead, the chatbot served only as a research assistant, helping Conyngham parse medical literature and identify potential treatment paths. University laboratories, specialized equipment, and months of expert labor were required – “not just a chatbot and a few prompts,” David Ascher, Professor of Biotechnology at the University of Queensland, noted.
Beyond ChatGPT, AlphaFold, Google’s protein structure model cited in some reports, has known limitations that further complicate its credited role. Ascher noted that while AlphaFold can generate structural hypotheses about proteins, it is not a turnkey system for designing cancer vaccines. Official guidance warns it is not validated for predicting mutation effects.
Similarly, Grok’s contribution is even harder to pin down. Conyngham posted on X that the final vaccine construct was designed by xAI’s chatbot, but what that means in practice, or what inputs the model received, remains unclear. No independent scientist has verified that claim.
“The ‘AI made this’ framing ignores this massive human effort, without which the AI’s output would have remained just text on a screen.”
Alvin Chan, Assistant Professor at Nanyang Technological University (via The Verge)
In broader terms, crediting software for outcomes that required extensive human expertise and institutional support reflects a recurring pattern in tech coverage – one that erases the labor of scientists who produced results. ChatGPT has been linked to deadly delusions and reinforcing false beliefs, yet its outputs continue to be presented uncritically.
In Rosie’s case, the path from chatbot query to injectable vaccine required a professor of RNA science, a canine immunotherapy specialist, genomic sequencing facilities, and a regulatory framework that took months to satisfy. None of that work was automated.
How the Hype Spread
Yet the gap between what experts describe and what the public heard is vast. Newsweek headlined Rosie’s story as a cure, declaring that an owner with no medical background had invented one. The New York Post wrote that ChatGPT coded a custom cancer vaccine.
At the same time, OpenAI president Greg Brockman shared the story on social media without questioning its accuracy.
How AI empowered Paul Conyngham to create a custom mRNA vaccine to cure his dog’s cancer when she had only months to live. The first personalized cancer vaccine designed for a dog: https://t.co/Outsl0VVMW pic.twitter.com/2uQn9bNA9t
— Greg Brockman (@gdb) March 14, 2026
Elon Musk amplified it and credited xAI’s Grok. Google DeepMind CEO Demis Hassabis shared it more cautiously.
Indeed, just the beginning
— Elon Musk (@elonmusk) March 15, 2026
First reported by The Australian, the story went viral with claims AI cured cancer, the kind of narrative tech executives have long sought. However, none of the executives who amplified the story noted that the treatment had not been shown to work, that AI had played only a supporting role, or that the vaccine was given alongside another therapy that may have been responsible for any improvement.
This amplification did not occur in a vacuum. OpenAI has separately been discussing AI in drug evaluation with the FDA, and millions of people globally ask health questions on ChatGPT Health every week.
Against that backdrop, uncritical amplification of an unverified cure story by the company’s own president carries particular weight. Brockman, who cofounded OpenAI and serves as its president, should have recognized that sharing an unverified medical claim about his company’s product would lend it undue credibility.
“[Rosie’s case] is better seen as an unusual, highly specific proof of possibility than a template ordinary people can readily reproduce.”
David Ascher, Professor of Biotechnology at University of Queensland (via The Verge)
Meanwhile, Conyngham’s X profile says “Ending Cancer for Dogs” and links to a Google form asking whether visitors are researchers or investors, raising questions about whether the viral story serves scientific or commercial interests.
Where mRNA Cancer Vaccines Stand
Beyond the hype cycle, legitimate mRNA cancer vaccine research is real but far from the viral narrative. Over 400 cancer vaccine trials have been initiated globally in the last three years, and Moderna and Merck’s experimental melanoma vaccine reduced death or recurrence risk by nearly 50% over five years.
Despite those promising results, more than 120 clinical trials have tested mRNA vaccines for cancer since the early 2000s, yet these treatments remain largely unproven in humans, let alone dogs. Personalized therapy remains hard to scale, as “personalized therapy is hard to do,” Mansoor Amiji, University Distinguished Professor at Northeastern University, noted.
Accordingly, AI is being used to identify tumor-specific antigens and improve vaccine efficacy, but that research proceeds through controlled clinical trials, not viral social media posts.
By comparison, legitimate mRNA cancer vaccine trials involve hundreds of patients, years of data collection, and regulatory oversight. Conyngham’s single-dog experiment, however compelling as a personal story, produced no publishable data and no peer-reviewed results, making it unsuitable as evidence for AI’s medical capabilities.
Scientists are performing tests to check Rosie’s immune response to the treatment. Separately, Conyngham told the New York Post he is developing a second vaccine for a tumor that did not respond.
Whether the first vaccine contributed to Rosie’s partial improvement, or whether the checkpoint inhibitor alone produced the effect, is unlikely to be answered without a controlled study.
Rosie’s case is real, but the narrative around it obscures more than it reveals about AI’s role in medicine. AI may be making scientific literature more accessible, but accessible information is not the same as accessible care.
OpenAI itself has acknowledged that millions discuss sensitive health topics every week, making uncritical amplification of unverified cure stories all the more consequential. For other pet owners and patients searching for answers online, the takeaway is sobering: without world-class researchers, genomic sequencing labs, and thousands of dollars in funding, a ChatGPT prompt will not produce a treatment.
