Agentic AI is set to transform user interactions with crypto wallets, especially concerning trading and payments. While AI and blockchain leaders suggest potential safety, it brings forth a new array of risks.
Recently, crypto exchange Coinbase unveiled its Payments MCP tool, enabling AI agents to access the same on-chain financial tools as humans.
Announcing Payments MCP, the easiest way for AI agents to get onchain via x402. 🚀
It allows LLM models like Claude, Gemini, and ChatGPT to access on-chain tools such as wallets, onramp, and payments without needing an API key. 🧵 pic.twitter.com/MSnIaecx0O
— Coinbase Developer Platform🛡️ (@CoinbaseDev) October 22, 2025
When used with an LLM like Claude, Gemini, and Codex, the tool enables autonomous access to crypto wallets and payment functionalities, as stated by the Coinbase Developer Platform here.
The AI agents powered by Payments MCP can handle payments, compute tasks, retrieve paywalled info, tip creators, and manage certain business operations through the x402 protocol, an open, web-native payment system for instant stablecoin payments, as indicated by Coinbase.
“This signifies a new era of agentic commerce where AI agents can operate within the global economy,” remarked the Coinbase Developer Platform.
Agentic AI in crypto is potentially safe
Aaron Ratcliff, attributions lead at blockchain intelligence firm Merkle Science, mentioned to Cointelegraph that from a security perspective, granting access to an AI agent introduces a layer of trust to an inherently trustless system.
While it can be secure if constructed properly, Ratcliff believes the onus of “safety” ultimately lies with the crypto user.
“Safe usage hinges on users’ understanding of prompts and the AI’s ability to pull blockchain data accurately. Additionally, safeguarding trading credentials is crucial; leaked credentials result in self-inflicted damage.”
AI in your portfolio may introduce additional risks
A recent survey from CoinGecko involving 2,632 crypto users revealed that the majority are open to AI managing their investments; 87% indicated they’d let AI agents oversee at least a tenth of their crypto portfolios.
Ratcliff pointed out some security threats that could be exploited if AI is involved in a person’s portfolio. Instruction or prompt injection could lead to system hijacking.
A man-in-the-middle attack, where a hacker positions themselves between communicating entities to steal data, could also misdirect trades.
“The AI might interact with scam tokens, miss honeypots or rug-pulls, or handle slippage poorly, risking users’ funds,” Ratcliff further commented.
“I’d need assurance that the AI could detect front-running, enforce slippage limits, identify scam tokens, and audit contracts in real time before executing a trade. It should also sandbox prompts, block injections, and prevent man-in-the-middle access.”
Moreover, Ratcliff noted that compliance gaps could lead to challenges, such as a lack of controls preventing AI from transmitting funds to sanctioned addresses or exchanges.
Stay vigilant even with AI safeguards
In conversation with Cointelegraph, Sean Ren, co-founder of Sahara AI, an AI-native blockchain platform, pointed out that Coinbase’s tool employs model context protocols, “which are the gold standard for security when configured properly.”
“They essentially function as a gatekeeper between the AI model and your wallet. The agent can only perform specific, authorized actions—like checking balances or setting up a payment for your confirmation—rather than autonomously shifting funds or modifying wallet settings,” he explained.
“These functions are restricted by design, ensuring that even if someone attempts to deceive the AI through prompt injection, it can’t complete transactions independently,” Ren added.
However, Ren cautioned that being safer does not equate to being infallible, and users need to remain alert regarding their AI agent’s activities concerning their portfolio.
“Users must stay vigilant, verify their approvals, and should never presume the agent is automatically making the correct decisions. Reviewing and authorizing transactions is still essential.”
The early stages for AI agents
Brian Huang, co-founder and CEO of Glider, a platform for AI-driven crypto portfolio management, shared with Cointelegraph that basic functions like sending, swapping, and lending are a sensible starting point for agents, though the field is still in its infancy.
Related: AI agents require crypto to function in financial markets: Coinbase executive
“These are straightforward tasks completed with a click—you wouldn’t ask ChatGPT to Venmo your friends, right? Many actions take longer with agents,” he noted.
“In contrast, agents act more like assistants; we’re all aware that DeFi is too intricate for general participation. These agents can aid users in getting onboarded and feeling supported throughout the procedure.”
Huang anticipates that more advanced actions, like portfolio management, rebalancing, and personalized financial advice, will emerge as more effective use cases.
“The customization that agents can offer, considering numerous variables, far surpasses what any human can provide,” he stated.
Magazine: How do the world’s major religions view Bitcoin and cryptocurrency?
