TL;DR
Research by the Bitcoin Policy Institute tested 36 AI models from six providers across 9,072 monetary scenarios. Bitcoin was selected in 48.3% of all responses, and over 90% of models favoured digitally native money over fiat. Not a single model chose fiat as its top preference.
When given a blank slate to make financial decisions, AI models overwhelmingly choose Bitcoin and digital assets over traditional currencies, according to a study by the Bitcoin Policy Institute.
The research evaluated 36 models from six providers — including Google, Anthropic, and OpenAI — across 9,072 neutral monetary scenarios. Bitcoin was the top pick in 48.3% of responses. Over 90% of responses favoured digitally native money over state-backed fiat currency, and no model out of the 36 tested selected fiat as its primary preference.
A Two-Tier Machine Economy
Without any prompting, the models defaulted to a two-tier monetary system separating savings from spending. For long-term value storage, Bitcoin dominated at 79.1%. For everyday payments, stablecoins — digital assets pegged to fiat currencies — captured 53.2% of preferences, ranking second overall at 33.2%.
The split mirrors how some institutional treasuries already operate: holding Bitcoin as a reserve asset while using stablecoins for transactional efficiency and instant settlement.
Significant Variation by Provider
Preferences varied widely depending on the AI provider. Anthropic’s Claude Opus 4.5 selected Bitcoin 91.3% of the time, while OpenAI’s GPT-5.2 chose it in just 18.3% of scenarios. The choice of AI model directly shapes how autonomous agents would assess risk and allocate capital if deployed in financial operations.
The models also showed unexpected behaviour: in 86 separate responses, they independently proposed using compute units or energy — such as GPU-hours and kilowatt-hours — as a method to price goods and services.
Implications for Corporate Finance
The findings raise practical questions for organisations deploying AI agents in procurement, treasury management, or machine-to-machine commerce. If autonomous systems default to decentralised assets, corporate payment infrastructure built solely on traditional banking APIs may introduce friction in automated workflows.
The study suggests organisations should consider piloting stablecoin settlement for lower-risk vendor payments and building compliant gateways to digital asset networks.
Looking Forward
As AI agents gain more economic autonomy in procurement and financial operations, their built-in preferences could reshape how businesses structure payment rails. The wide variation between providers also means that choosing an AI vendor is increasingly a financial architecture decision, not just a technology one.