Opinion: AI agents may betray the core promises of the crypto world, as their "black box" systems cannot be verified or audited

BlockBeats News, October 20, according to Forbes, by 2025, autonomous AI agents have become one of the hottest narratives in the crypto industry, expanding overnight from an experimental novelty to a market worth $13.5 billions. An AI agent called "Truth Terminal" once convinced well-known venture capitalist Marc Andreessen to donate $50,000, which drove the GOAT token's market cap to soar to $1.2 billions.

Currently, on the Virtuals Protocol platform alone, there are more than 11,000 AI agents running, executing trades and managing portfolios with minimal human intervention.
But there is a problem that almost no one is willing to face. These AI agents were originally designed to improve the efficiency of DeFi, yet they themselves are often highly centralized. The vast majority rely on closed-source models from companies such as OpenAI and Anthropic, resulting in a centralized monopoly at the expense of user data and trading flow.

In an industry built on transparency, AI agents simultaneously represent the most market-fit product the crypto world has seen so far—and the most severe ideological contradiction. The question is no longer "Will AI agents reshape the crypto industry," but rather—they already are. Security researchers warn that many AI agents deployed on blockchain networks use unaudited smart contracts, and most delegate decision-making processes to centralized AI services.

When an agent executes a $100,000 DeFi strategy, the real decision-making and reasoning actually takes place on servers owned by OpenAI or Google—these "black box" systems cannot be audited or verified by anyone.