Making scents: could AI help perfumers take pressure off endangered plants?

by | Oct 10, 2026 | Science

Making scents: could AI help perfumers take pressure off endangered plants?

The fragrance industry is increasingly turning to artificial intelligence and biotechnology to reproduce scents and aromatic compounds derived from endangered plant species, potentially reducing the environmental impact of harvesting rare botanicals.

One of the primary concerns driving this research is the depletion of wild Aquilaria trees, which produce agarwood or oudh, a highly valuable fragrance ingredient obtained when the trees are damaged. Illegal harvesting for this compound has devastated wild populations. Companies like Osmo, a Google Brain spin-off that has secured $130 million in funding, are developing AI systems capable of predicting and recreating scents. The company claims to have reproduced the scent of a rare tree using a bark sample the size of a finger, with spectroscopy and artificial intelligence reconstructing the chemical formula. BioHarvest Sciences has announced a different approach, creating a stable cell culture from an endangered southeast Asian plant, enabling production of its fragrance compounds without cultivating the original source. The company anticipates producing up to 20 tonnes of the fragrance composition, with limited production potentially beginning next year.

However, conservation experts and perfumers have expressed skepticism about these claims. Mandy Aftel, a pioneering natural perfumer, characterizes the technology as greenwashing, arguing that synthetic alternatives do not actually preserve endangered species and may not reduce demand for wild-sourced materials. Rhian Smith from the Royal Botanic Gardens, Kew, notes that despite potential AI alternatives to traditional oudh, wild-sourced material remains highly valued by consumers, meaning technological advances may not fully alleviate pressure on wild populations.

The technological advances themselves represent genuine innovation in scent analysis and recreation. Osmo claims to have trained its AI on more than 5 million human responses to smells, enabling prediction of how molecular combinations will be perceived. Other approaches include reconstructing genes from preserved flowers to produce fragrance compounds and collecting living microbes from clouds as bases for new perfumes. Yet researchers acknowledge significant limitations, noting that a flower’s scent can contain hundreds of molecules that vary based on plant part and growing conditions, making complete recreation impossible.

Experts emphasize that the recreation of scent remains fundamentally incomplete, capturing only fragments of complex natural aromas while unable to replicate the living organism or environmental conditions that produce original scents. The fundamental question of whether technological substitutes genuinely serve conservation goals remains contested within both the perfume and conservation communities.

Article Attribution | Read More at Article Source

Article summary produced by Claude AI