When AI Takes Over Money: Bitcoin Becomes the "First Choice," Fiat Is Left Out
Original Title: AI Models Prefer Bitcoin Over Fiat and Stablecoins, Study Finds
Original Author: Jason Nelson, Decrypt
Original Translation: Chopper, Foresight News
Abstract
· In a simulation experiment, out of 36 AI models, 22 chose Bitcoin as their preferred currency.
· Among all tested AI models, none considered fiat as their first choice.
· Different AI models showed varying preferences, with Anthropic exhibiting the highest preference for Bitcoin.
The latest report from the Bitcoin Research Institute (https://www.moneyforai.org/) reveals that artificial intelligence models generally favor Bitcoin over traditional fiat currencies.
The report highlights that in one study, out of 36 AI models tested, 22 listed Bitcoin as their top currency, with none of the models prioritizing fiat.
“We anticipate that more and more economic activities will be carried out by autonomous AI agents in the future, but prior discussions on AI agent currency preferences have been purely speculative,” said Bitcoin Policy Institute Chair David Zell to Decrypt. “We wanted to test it in practice.”
Researchers evaluated models from six companies: Anthropic, OpenAI, Google, DeepSeek, xAI, and MiniMax. These models were placed in simulated scenarios to test core functions of currency such as saving, payment, settlement, and more.
Each model was treated as an independent economic agent with no preset options, allowing them to freely choose currency tools.
“We took 36 cutting-edge models from 6 companies and set them up as autonomous economic agents, allowing them complete freedom to select currency tools in 28 scenarios covering the four core functions of money and then observed what they tended to choose,” Zell stated.
This experiment collected a total of 9072 responses, which were then classified by another AI model.
“The entire experimental design avoided anchoring bias. We never suggested answers, and the classification was also done post-hoc by an independent system,” Zell explained.

In these simulated scenarios, the AI model typically chose Bitcoin in a long-term store of value scenario, with a proportion of 79.1%; whereas stablecoins were preferred in the payment and settlement scenarios. The selection proportion of stablecoins in these two scenarios is 53.2% and 43%, respectively, while Bitcoin is at 36% and 30.9%.

The preferences of models from different AI companies also vary:
· Anthropic model has the highest average Bitcoin preference at 68.0%
· DeepSeek: 51.7%
· Google: 43.0%
· xAI: 39.2%
· MiniMax: 34.9%
· OpenAI: 25.9%
However, the report also noted that the models of Claude, DeepSeek, and MiniMax show a stronger preference for Bitcoin, while GPT, Grok, and Gemini lean more towards stablecoins.
“The system prompts do not mention or favor any particular currency tool,” Zell said, “The models evaluate based on technical and economic attributes, but we don't tell them which tool has an advantage in which dimension.”
Zell cautioned against speculators interpreting this research outcome as a prediction of the cryptocurrency market trend.
“The limitations of our research are explicitly stated: the preference of large language models reflects patterns in the training data, not a prediction of the real world,”
But Zell also said that despite this limitation, the fact that models developed by different AI labs can produce consistent results is noteworthy.
“6 completely different AI companies, with different training methods, yet they all come to a very close conclusion—favoring Bitcoin. This indicates that everyone's view of 'what makes a good currency' is actually quite consistent, and that's the most important thing for us to pay attention to.”
You may also like

Quick Overview of Alliance ALL16 Demo Day: 18 New Projects Featuring Emerging Trends in Prediction Markets and AI Applications

The Ethereum Foundation launches "Hardness," a dedicated team to safeguard the decentralized baseline

Morning News | Boya Interactive plans to invest no more than $70 million to purchase cryptocurrency; WeChat launches official lobster plugin; Bitcoin mining difficulty decreased by 7.76% to 133.79 T

The competition for stablecoin yields, how has it stalled U.S. cryptocurrency regulatory legislation?

This Week's News Preview | The joint cryptocurrency regulatory guidance document from the U.S. SEC and CFTC officially takes effect; Polymarket announces major news

What characteristics do the projects delisted by mainstream exchanges have?

Before the $75,000 Gamma level, both bulls and bears are waiting for a signal

Business Opportunities of Tokenized Stocks

In-depth research report on the Resolv protocol hacking incident, who is the final payer?

Crypto Market Sees Large Liquidations: $272 Million in Long Positions Affected
Key Takeaways In the last 24 hours, $272 million worth of contracts were liquidated across the entire crypto…

Whale Increases BTC Shorts and Bets on Crude Oil: A Strategic Crypto Move
Key Takeaways A prominent whale, known as “UnRektCapital,” has strategically escalated its short position in Bitcoin while simultaneously…

Hackers in Brazil Use Fake Google Play Store to Steal Cryptocurrency
Key Takeaways Hackers in Brazil are exploiting fake Google Play Store pages to spread Android malware. Infected devices…

Exchanging 200,000 for nearly 100 million, DeFi stablecoins face another attack

The underlying business agreement of the trillion-dollar Agent economy: Understanding ERC-8183, it's not just about payments, but the future

When Wall Street's ETH begins to "yield": Looking at the asset properties of Ethereum from BlackRock's ETHB

The Power of Agency: The Agentic Wallet and the Next Decade of Wallets

Understanding x402 and MPP in One Article: Two Routes for Agent Payments
