{"paper_id":"24cd5c46-1567-4931-9e83-e4b6abd00670","body_text":"This is a Preprint and has not been peer reviewed. This is version 2 of this Preprint.\nYou must log in to post a comment.\nThere are no comments or no comments have been made public for this article.\nThis is a Preprint and has not been peer reviewed. This is version 2 of this Preprint.\nAdd a Comment\nYou must log in to post a comment.\nComments\nThere are no comments or no comments have been made public for this article.\nBiodiversity credits are being pushed as a means to fund nature conservation. Much of the debate around credits has concerned additionality, leakage, and permanence, and the extent to which biodiversity can be captured in an individual unit. As AI models continue to develop, however, technology could create a new kind of loss-of-control problem for biodiversity credits. In this Perspective, we express a concern that agentic AI could lead to the development of high-risk autonomous credits, where nation states cede control of nature restoration to the price of digital assets that are paid out with no human in the loop. We define an autonomous credit as a financial asset for nature quantified remotely through an agentic AI combined with a spatial finance model and real-time programmatic payment. Such autonomous credits do not yet exist, but appear to be in development. We highlight three technological step changes that signal autonomous credits are on the horizon, and then suggest three approaches that might help mitigate loss-of-control and maximise the benefits of credit technology. Given the development of agentic AI, where autonomous credits are connected to the critical infrastructure of food production and finance, these credits will need guardrails.\nhttps://doi.org/10.32942/X2XM2Z\nBiodiversity, Life Sciences, Population Biology\nbiodiversity credits, agentic AI, spatial finance, programmable payment\nPublished: 2026-04-21 21:04\nLast Updated: 2026-05-09 09:19\nCC BY Attribution 4.0 International\nConflict of interest statement:\nNone\nData and Code Availability Statement:\nNot applicable\nLanguage:\nEnglish","source_license":"CC-BY-4.0","license_restricted":false}