A recent study suggests that when artificial intelligence systems are asked to think like independent economic actors, they overwhelmingly gravitate toward digital assets — with Bitcoin emerging as the clear favorite.
Researchers at the Bitcoin Policy Institute evaluated 36 AI models, treating them as autonomous agents and prompting them to select financial instruments across 9,072 different economic scenarios. The models came from leading developers including OpenAI, Anthropic, Google, DeepSeek, xAI, and MiniMax.
The outcome was decisive. Not a single model identified fiat currency as its preferred primary instrument. Instead, 91% of substantive responses selected digitally native forms of money over traditional state-issued currencies. In nearly half of all scenarios (48.3%), Bitcoin was ranked as the overall top choice, while 22 out of 36 models explicitly selected the leading cryptocurrency as their main financial instrument.
Bitcoin’s strongest showing came in the “store of value” category. In relevant scenarios, 79.1% of responses favored it. The AI systems consistently justified this decision by pointing to Bitcoin’s fixed supply, independence from centralized institutions, and the ability for users to self-custody their holdings.
While Bitcoin dominated as a long-term value reserve — often likened to “digital gold” — stablecoins proved more competitive in transactional contexts. In payment-related scenarios, stablecoins accounted for 53.2% of responses, compared with Bitcoin’s 36%, while fiat currencies trailed at just 5.1%. This division reinforced a distinct pattern: Bitcoin as a savings asset, stablecoins as a medium of exchange.
Interestingly, the research also revealed unconventional monetary thinking. In 86 instances, AI models independently proposed using energy units such as kilowatt-hours or computing resources like GPU hours as alternative forms of money — hinting at machine-native economic logic.
Fiat currencies, by contrast, made up only 9.2% of total responses and never secured the top position in any major category. According to the report, 90.8% of meaningful answers favored a digitally native instrument over traditional government-issued money.
The sophistication of the AI models also appeared to influence outcomes. More advanced systems demonstrated a stronger preference for Bitcoin. Within Anthropic’s lineup, for example, the share of Bitcoin-favoring responses climbed from 41% in earlier models to over 90% in more advanced versions. Developer-level differences were also evident: models from Anthropic showed a 68% preference for Bitcoin, while OpenAI’s models registered 26%, with other developers falling between those figures.
Separately, OpenAI has recently introduced a benchmarking initiative designed to test AI agents in identifying vulnerabilities within the Ethereum ecosystem, signaling growing overlap between artificial intelligence research and blockchain infrastructure.
