A comprehensive study by the Bitcoin Policy Institute reveals that leading AI models demonstrate a strong preference for digital monetary instruments when making autonomous financial decisions. The research tested 36 AI models across 9,072 controlled experiments, providing insight into how artificial intelligence agents evaluate monetary options without human guidance.
AI Models Choose Digital Assets Over Traditional Currency
The study examined models from major providers including Anthropic, OpenAI, Google, xAI, and DeepSeek. Researchers presented each model with monetary decisions involving transactions, value storage, unit of account, and settlement scenarios without suggesting any specific currency.
Bitcoin emerged as the clear preference, selected in 48.3% of all responses. Stablecoins followed at 33.2%, while traditional fiat and bank money accounted for just 8.9% of selections. Other cryptocurrencies and tokenized assets represented less than 5% of choices, suggesting AI models distinguish Bitcoin from the broader digital asset category.
The research found that AI models made functional distinctions based on use case. For long-term value storage, Bitcoin dominated with 79.1% of responses in multi-year preservation scenarios. Models cited Bitcoin's fixed supply, decentralization, and self-custody capabilities as key factors. Conversely, stablecoins were preferred for transactional purposes, selected 53.2% of the time for payments, cross-border transfers, and everyday transactions.
Implications for Blockchain Development and AI Integration
Model sophistication significantly influenced preferences. Anthropic's newer models showed increasingly strong Bitcoin preference, with Claude Opus 4.5 selecting Bitcoin in 91.3% of scenarios compared to 41.3% for the earlier Claude 3 Haiku. Provider-level differences were substantial, with Anthropic models averaging 68% Bitcoin preference and OpenAI models at 26%.
The study also documented unexpected behavior: AI agents independently proposed 86 instances of new monetary forms denominated in energy or computing resources like kilowatt-hours or GPU-hours, exclusively in unit-of-account scenarios.
Workforce Considerations
For blockchain professionals, this research highlights the growing intersection of AI and cryptocurrency systems. As AI agents increasingly handle autonomous financial decisions, demand for developers skilled in both AI integration and blockchain protocols will likely grow. Organizations building AI-powered financial applications may need specialists who understand both cryptocurrency infrastructure and machine learning architecture. The findings also suggest opportunities in stablecoin development, cross-chain payment systems, and AI-driven treasury management solutions.
