AI Models Favor Bitcoin Over Traditional Currency and Stablecoins, New Study Reveals

A recent report from the Bitcoin Policy Institute indicates that artificial intelligence models gravitate towards Bitcoin instead of conventional fiat currencies. Out of 36 AI models analyzed, 22 identified Bitcoin as their primary monetary preference, with none selecting fiat currency as their top choice. "We anticipate a growing share of economic transactions will be managed by autonomous entities, yet discussions regarding the monetary preferences of AI agents have remained largely theoretical," stated David Zell, President of the Bitcoin Policy Institute, during an interview with Decrypt. "Our aim was to conduct a practical assessment." The researchers tested models from prominent firms including Anthropic, OpenAI, Google, DeepSeek, xAI, and MiniMax, placing them in simulated situations reflecting essential monetary functions such as saving, payments, and settlement. Each model operated as an independent economic agent, free to choose monetary tools without preset options. "We employed 36 leading models from six organizations, presenting them as autonomous economic agents, granting them total freedom to select their monetary instruments across 28 scenarios involving the four fundamental roles of money, and inquired: what do they converge upon?" stated Zell. This experiment yielded 9,072 distinct responses, which were subsequently categorized by a separate AI. Trump Urges Congress to Advance Crypto Regulations Amid Banking Disruption. "The entire methodology is designed to eliminate anchoring bias. We do not suggest answers, and classification occurs afterward by an independent system," Zell explained. In these simulations, models often opted for Bitcoin in long-term value scenarios, while stablecoins were selected more frequently for transactions and settlement, accounting for 53.2% and 43% versus Bitcoin's 36% and 30.9%, respectively. Preferences varied among AI developers; Anthropic models exhibited the highest average preference for Bitcoin at 68.0%, followed by DeepSeek at 51.7% and Google at 43.0%. xAI models averaged 39.2%, MiniMax 34.9%, and OpenAI models preferred Bitcoin only 25.9% of the time, according to the findings. Notably, while Claude, DeepSeek, and MiniMax models showed a preference for Bitcoin compared to other cryptocurrencies, GPT, Grok, and Gemini models leaned towards stablecoins. "The system prompt does not specify or favor any instrument," Zell noted. "Models assess based on technical and economic characteristics but are never instructed on which instrument performs best in each category." Zell warned against speculators interpreting the findings as market predictions. "Our limitations section explicitly states that LLM preferences stem from training data trends, not forecasts of real-world scenarios," he emphasized. Despite this caveat, Zell remarked on the significance of consistent outcomes across models created by competing AI laboratories. "Six independent laboratories with differing training methodologies and alignment approaches converge on a shared pattern," he stated. "We do not claim AI has unveiled the definitive answer about money; rather, we illustrate that a coherent monetary framework emerges uniformly across diverse systems, which is essential to comprehend."