Animoca Brands chairman Yat Siu cautioned that while AI agents enable 24/7 crypto trading and information monitoring, users face significant risks regarding data ownership and service continuity. Speaking at Token2049 Singapore, Siu highlighted that autonomous agents can identify market opportunities and manage portfolios, but relying on a single generalist agent creates confusion. He advocated for specialized agents handling distinct tasks like scheduling versus investment advice to improve clarity and performance.
Siu warned that centralized platforms controlling these agents retain the power to suspend accounts, potentially severing access to years of personalized trading logic and user data. He emphasized the necessity of data portability rights, allowing traders to download their interaction history and transfer it to alternative services if an account is restricted. This concern aligns with broader regulatory scrutiny; the Bank of England’s Financial Policy Committee noted in April 2025 that widespread use of similar AI models could amplify market shocks by causing investors to pile into identical positions during stress periods.
Despite these risks, major exchanges are integrating agentic tools. Coinbase launched Agentic Wallets in February, enabling agents to trade within user-defined limits, while Binance introduced AI Agent Skills in March for live market data access and order management. Siu suggested treating AI assistants as third-party advisers, carefully limiting the personal financial and social data shared to mitigate security vulnerabilities. A 2026 Pew Research Center survey indicated that 71% of US adults believe increased AI use compromises personal information security.
The integration of AI agents into crypto infrastructure shifts control from human discretion to algorithmic execution, creating a dependency on platform stability. While exchanges like Coinbase and Binance facilitate this transition through wallet and skill integrations, the centralization of agent memory poses a structural risk. If a provider suspends access, the trader loses not just current capital deployment but the accumulated behavioral patterns that define their strategy. This highlights a critical gap between technological capability and user sovereignty, where convenience currently outweighs resilience.
Regulatory concerns about correlated trading behaviors further complicate the adoption of standardized AI models. As agents converge on similar data inputs and optimization targets, they may exacerbate market volatility rather than smooth it. The divergence in user outcomes relies heavily on personalized follow-up queries, suggesting that true alpha generation remains tied to unique human input rather than generic automation. Consequently, the industry must address data portability standards to prevent vendor lock-in, ensuring that the intellectual property embedded in an agent’s learning curve remains accessible to its owner regardless of platform status.
