AI scent recreation targets endangered plants for luxury perfumes

New AI tools could reduce pressure on rare plants used for high-end perfumes by creating synthetic scents from small samples.

Osmo, a Google Brain spin-off, has raised $130 million and trained its AI on over 5 million human responses to smells. The system recreated the scent of a single rare tree using a finger-sized bark sample. BioHarvest Sciences developed a stable cell culture for an endangered Southeast Asian fragrance plant and announced an agreement to produce 20 tonnes of fragrance composition starting next year. BioHarvest’s CEO stated human demand for such compounds is 'effectively unlimited' compared to biological supply, while wild Aquilaria sinensis trees—used for illegal oudh harvesting—remain critically endangered.

The technology leverages AI predictions of molecular combinations and synthetic biology to bypass direct plant harvesting. However, Osmo admits conservation isn’t an explicit priority, and Mandy Aftel of the Aftel Archive called synthetic recreation of endangered plant scents 'greenwashing.' Rhian Smith of Kew noted wild-sourced oudh remains highly prized despite synthetic alternatives.

This approach could reduce immediate pressure on specific endangered species for luxury perfumes, potentially freeing up biodiversity for other uses. But it addresses only narrow commercial contexts: BioHarvest’s 20-tonne target is limited to fragrance production, and wild populations remain devastated by illegal harvesting. Osmo’s AI cannot fully replicate natural scents due to hundreds of variable molecules and environmental factors. The real conservation impact depends on scaling beyond single species and addressing illegal trade—not just scent replication.

Source: Guardian