University of Melbourne researchers have utilized an artificial intelligence tool to highlight critical animal welfare concerns prevalent in the United Kingdom's greyhound racing sector. This development has reignited global discussions regarding potential bans on the sport.

The study, published in Frontiers in Animal Welfare and Policy, marks the first instance of employing AI agents to systematically analyze data from over 31,000 licensed greyhounds that participated in more than 1.26 million races in the UK between early 2022 and March 2026. The aim was to gain a clearer understanding of the lives and well-being of these animals.

AI Uncovers Hidden Data

Dr. Mia Cobb, the lead researcher and an animal welfare scientist, explained that the AI system was instrumental in identifying concerning details previously obscured within publicly accessible online records. The UK greyhound racing industry generally portrays itself as transparent, sharing certain welfare figures annually. However, Dr. Cobb noted that these published statistics do not offer a true picture of a greyhound's lived experience.

The research employed agentic AI, a type of software that automates repetitive tasks under human oversight. This allowed Dr. Cobb to aggregate information from multiple public websites, uncovering animal welfare insights with unprecedented speed and scale, a task that would have previously taken a research team many months to accomplish.

Fatality Rate Discrepancies

One of the initial alarming findings by the AI tool was the industry's reported stable rate of on-track fatalities occurring concurrently with a rise in actual deaths and a decline in race numbers. This discrepancy prompted further investigation.

"We first noticed the industry reported a stable rate of on track fatalities when the number of deaths was rising and race numbers were falling, which didn't make sense," Cobb said. "Subsequent analysis revealed that the on track fatality rate for racing greyhounds had actually increased by 30 percent between 2022 and 2024. "By rounding fatality rates to two decimal places, the regulator's reporting over the past three years concealed the fact that more dogs were dying even as race starts declined. "The UK's greyhound welfare strategy cannot claim improved canine safety, with more than two dogs dying on track every week since its introduction in 2022. This led us to use AI agents to explore what else was going on."

Post-Race Welfare Concerns

Dr. Simon Coghlan, a co-researcher and the deputy director of the university's Centre for AI and Digital Ethics, expressed concerns that the reported on-track fatalities might only represent a fraction of the total deaths. Public registries, he pointed out, do not record fatalities that occur after races due to injuries sustained during competition.

"Our research highlights a major visibility issue," Coghlan stated. "There is no public registry that accounts for individual dogs' post racing welfare. However, we know the Greyhound Board of Great Britain holds unpublished data on retirement destinations, career ending reasons, as well as individual level injury and euthanasia figures linked to specific racetracks. "Restricting access to this information allows the industry to go unchecked regarding its accountability to animal welfare claims."

The study also revealed that the high turnover rate of greyhounds has remained unchanged, with approximately 40 percent of racing dogs exiting the sport annually. The industry sustains its operations through a consistent influx of new greyhounds, predominantly from Ireland, which constitute 85 percent of those racing in the UK. This reliance means nearly half of the greyhounds bred annually in the Republic of Ireland are destined for UK tracks.

Global Regulatory Landscape

Further analysis by the AI identified varying rates of harmful incidents, such as crashes and falls, across different racing venues. Dr. Cobb noted that the typical racing career for a greyhound is less than a year, and some tracks exhibit a higher frequency of these dangerous events. This granular data has not been disclosed by the regulator, even when specifically requested by a Welsh parliamentary committee considering a ban.

The findings emerge as several countries and regions are re-evaluating their stance on greyhound racing. Scotland and Wales have announced plans for bans or phase-outs. New Zealand was set to prohibit the sport by July 31 of this year, after initially announcing a phase-out in December 2024. In Australia, the ACT has already banned the sport, Tasmania plans to phase it out by mid-2029, South Australia faces a July 2026 reform deadline or a potential statewide ban, and Western Australia's Parliament is conducting a formal inquiry into the matter.

Transparency and Future Applications

Dr. Suzie Fowler, RSPCA Australia Chief Science Officer, commented on the persistent animal welfare issues within the Australian greyhound industry, including overbreeding, inadequate living conditions, high injury and mortality risks, and a lack of transparency concerning retired dogs' fates. She highlighted the importance of transparency and publicly accessible data regarding animal welfare outcomes.

"For decades, the RSPCA has called for many industries, including greyhound racing, to improve their transparency and provide more publicly available and visible data regarding animal welfare outcomes," Fowler said. "This new methodology does the heavy lifting of compiling data that would otherwise take a team months to complete, but we must remember it is still only as helpful as the data made publicly available by industry. "Tools like this will help identify key animal welfare concerns and opportunities for improvements by industry and government bodies and could be used to assess effectiveness of changes to improve welfare."

The research team hopes their study will influence policymakers globally as they review greyhound racing regulations. "Industry transparency should mean genuinely showing the public how dogs are doing. This research shows that disclosure does not necessarily produce visibility. We hope the greater visibility the AI agents offered will enable better accountability to meet public expectations," Dr. Cobb concluded. She also expressed hope that the novel AI research method could be adapted for use by other researchers, regulators, welfare scientists, and journalists studying animal-reliant industries with publicly available data, thereby fostering greater understanding of how human decisions impact animals.