The algorithm will see you now: Why AI is the nail in the coffin for animal medical testing

Every 24 hours, between 100 million and 200 million animals are dying cruel deaths in laboratories worldwide. Mice, rats, rabbits, dogs and primates are used in the name of science and progress. We’ve been told for years that their suffering is “necessary” to keep medicines or cosmetics safe. But this is just wrong. The future of medical testing isn’t cowering in fear in a cage. It’s written in code.

Artificial intelligence models are now able to predict drug toxicity, organ damage and chemical hazards with accuracy that matches (and often exceeds) what experiments on non-human animals deliver. The moral case against animal testing has always been clear and now the scientific case has collapsed too.

Machine learning models trained on molecular structures and biological data now predict acute oral toxicity with accuracy rates between 80% and 92% (Allen et al, 2019). AI systems analysing heart tissue data forecast six major types of drug-induced cardiac damage with 79% accuracy for known drugs (Mamoshina et al, 2020). For skin allergies, one of the most common safety tests, a consensus AI model approved by the Organization for Economic Cooperation and Development (OECD) achieves 80% accuracy, eliminating the need for certain animal assays entirely (Imamura et al, 2026).

Traditional animal tests have always been crude proxies: a rat is not a human, a beagle’s liver doesn’t process drugs the way ours does, and a rabbit’s eye tells us very little about how a chemical will affect a person’s eyes. The so-called “gold standard” has always been fool’s gold.

What AI offers is the ability to learn from vast datasets of human biology, real-world drug outcomes, and molecular interactions at scales no laboratory could replicate. When researchers at the forefront of computational toxicology built Tox-GAN (a generative AI system) it produced synthetic liver tissue data so realistic it matched actual biological samples with 99.7% similarity (Chen et al, 2022). No non-human animals were required.

AI, machine learning, organ-on-chip systems and computational biology have opened the door to a future where medical progress doesn’t require a mountain of animal corpses.

For years, defenders of non-human animal testing have hidden behind regulatory requirements, claiming that laws “demand” these experiments. Yet the OECD (the body that sets international testing standards) now explicitly accepts AI-driven approaches for skin sensitisation testing (Imamura et al, 2026). The US Food and Drug Administration, once the world’s most conservative drug regulator, released a roadmap in 2025 supporting alternatives that “strategically reduce animal use while increasing human relevance” (Dinç et al, 2025). Even the European Union is embedding computational methods into its chemical safety frameworks. Virtual control groups (AI-generated comparison data that replaces live control animals in experiments) have been validated across 20 studies, showing they can cut animal use substantially without compromising scientific rigour (Lofti et al, 2026). Explainable AI platforms like KidneyTox now let chemists predict kidney damage from drug candidates and understand why a molecule is toxic, allowing for better design, right from the start (Amin et al, 2026).

Why Africa must lead, not follow

For African nations, this technological shift should be seen as an opportunity. Our continent has long been positioned as a passive recipient of pharmaceutical research from the Global North, a testing ground for drugs developed elsewhere, with little say in how those drugs are evaluated. And we’ve also inherited regulatory frameworks built on animal testing not because they’re optimal, but because they’re traditional.

Africa can leapfrog outdated models the same way we bypassed landlines for mobile phones. By investing in AI-driven toxicology platforms, training computational biologists and advocating for regulatory acceptance of non-animal methods, we can build a medical testing infrastructure that is more ethical, more accurate and more aligned with 21st century science. South Africa, with its growing tech sector and research universities, is uniquely positioned to lead this transition on the continent.

Animal facilities are expensive to build and maintain. Breeding, housing and caring for laboratory animals drains resources that could fund computational infrastructure, data science training and human-relevant research. Once they are developed, AI models scale at near-zero marginal cost. A single validated algorithm can be deployed across dozens of institutions, democratising access to cutting-edge safety testing in ways animal labs never could (Durai et al, 2026).

Will we cling to the old ways, perpetuating cruelty out of inertia? Or will we embrace the tools that make compassion and scientific rigour compatible?

But more than the science and the economics, we need to keep in mind that this is really about: suffering. Every mouse injected with a test compound, every rabbit with chemicals dripped into its eyes, every beagle forced to inhale noxious gases and every primate subjected to invasive neurological experiments is an individual capable of pain, fear and distress. The fact that violence against non-human animals has been normalised doesn’t make it acceptable.

The philosopher Peter Singer once wrote that the question is not whether animals can reason or talk, but whether they can suffer. The answer has always been yes. The question now is whether we will continue to inflict that suffering. Science has given us an exit. AI, machine learning, organ-on-chip systems and computational biology have opened the door to a future where medical progress doesn’t require a mountain of animal corpses. Walking through that door requires political will, regulatory courage and a willingness to challenge entrenched ideas and interests: the breeders, the contract labs and the researchers whose careers are built on animal models.

Will we cling to the old ways, perpetuating cruelty out of inertia? Or will we embrace the tools that make compassion and scientific rigour compatible? Our government should mandate that all publicly funded research explore non-animal alternatives. Our universities should train the next generation of toxicologists in computational methods. Our regulatory agencies should align with international bodies moving towards AI-driven safety assessment, ensuring African patients benefit from human-relevant testing. And each of us, as citizens, consumers and moral agents, should demand better. We should ask pharmaceutical and cosmetics companies what they’re doing to phase out animal testing. We should support organisations that are developing and validating alternatives. We should buy products that are certified cruelty free. We should refuse to accept “that’s how it’s always been done” as justification for suffering.

The algorithm will see you now. It’s time we let the animals go. DM

Catherine Botha is professor of art, culture and technology in the University of Johannesburg’s philosophy department.