When Encryption Meets AI: Three Questions to Understand the Future of the Industry

Author: Miles Deutscher Source: X, @milesdeutscher Translation: Shan Ouba, Jinse Finance

In the crypto space, artificial intelligence is the vertical that excites me the most. But most people only see AI as a buzzword in the Web3 space.

Today, I want to unveil the mystery and delve into the true intersection of the two—intersections that can not only change the industry landscape but also hold enormous upside potential. I believe that the combination of cryptocurrency and artificial intelligence will transform the way the crypto market operates through real-world use cases, while simultaneously addressing key issues in the field of artificial intelligence.

This article will be divided into three core parts: Part 1: Practical Use Cases of Artificial Intelligence and Web3 Part 2: How Web3 Drives the Development of Artificial Intelligence Part 3: Potential Risks

Part 1: Practical Use Cases of Artificial Intelligence and Web3

For the average person, this part may be the most valuable because it delves into the methods of practically applying artificial intelligence in Web3.

1. Transaction Management

One ​​of the most interesting verticals in applying artificial intelligence in Web3 is using trained AI agents for transaction management and execution. For example, future trading may not require manual operation; instead, a "personal assistant" (i.e., an AI agent) can be deployed to handle transaction execution, position management, and other operations on your behalf.

These intelligent agents can not only execute transactions and manage investment portfolios on your behalf (some protocols are already developing related functions), but also achieve fully on-chain interaction. @HeyAnonai is a good example. Through its artificial intelligence protocol, you can complete operations such as transactions, cross-chain transfers, and on-chain interactions simply by using natural language commands. 2. Large Language Models (LLMs) in Web3 Many people believe that using large language models in Web3 is simply sending instructions to ChatGPT, but the actual applications are far more complex. Large language models can be viewed as an interface layer between humans and protocols—meaning that people can obtain data through natural language.

Large Language Models in the Web3 Domain

From a practical application perspective, with the widespread adoption of large language models, coupled with more protocols developing large language models specifically for training on Web3 data, significant barriers to accessing market information will be eliminated. Imagine: if you could simply input a command to access top-tier data and information 24/7, how much would your trading success rate increase?

3. Security and Privacy Protection

Artificial intelligence can trigger instant alerts within seconds, far exceeding human speed.

Artificial intelligence models analyze on-chain transaction sequences to identify typical attack behaviors, thereby triggering these immediate alerts. In this field, the core advantage of AI models over humans lies in their powerful pattern recognition capabilities. For ordinary people, this means fewer attack incidents and improved overall security of smart contract interactions. Part Two: How Web3 is Driving the Development of Artificial Intelligence In this section, I will delve into the latest content from a16z's "2025 Crypto Industry Status Report". The report points out that Web3/cryptocurrency can solve four key problems in the field of artificial intelligence: Validating human and AI activities; Achieving economic interaction between AI agents; Promoting standardized licensing of intellectual property rights; Ensuring fairness and openness in artificial intelligence. While solving these AI challenges is no easy task, cryptocurrencies have proven to possess a high-quality infrastructure for achieving these goals. We will analyze them one by one below: 1. Verifying Human and AI Activities While AI is still in its early stages (compared to its future potential), the problem of distinguishing between human and AI activities is already quite prominent. Cryptocurrencies can address this issue in three main directions: Proof-of-Humanity (PoH) Systems: Verifying that an operation is performed by a real human without revealing their identity. Protocols such as Worldcoin are already advancing this direction on a large scale. On-Chain Signature Activities: When an AI agent performs operations on behalf of a user (as described in Part 1), these operations can be verified through cryptographic signatures. On-chain accountability: Every action taken by an AI agent (whether executing a transaction, voting in a DAO, or modifying data) can be recorded, verified, and traced on-chain. 2. Enabling economic interaction of AI agents. As emphasized by a16z, to fully unleash the potential of AI agents, they must be equipped with economic interaction capabilities. This is precisely where x402 (and other infrastructure) comes in. As @suhailkakar explained, simply put, x402 is like adding a wallet function to the internet. It's a new network standard that allows websites to implement an interactive logic of "paying a small fee before providing data." x402 is just one example of how cryptocurrency empowers intelligent agent economic interactions. In addition to the infographic created by Suhail, I have also compiled an x402 introductory guide:

x402 - Easy-to-Understand Version.

3. Promoting the Standardized Licensing of Intellectual Property

Ensuring that owners possess legitimate intellectual property rights is key to the fair application of artificial intelligence technology. By putting these intellectual property rights on the blockchain, we can verify the compliance of their use and embed licensing terms into smart contracts. @campnetworkxyz is a protocol I have high hopes for in this field. Its protocol allows anyone to own and profit from intellectual property rights. Refer to "The Core Concepts of the Camp Protocol" for more information.

4. Ensuring Fairness and Openness in Artificial Intelligence Finally, applying Web3 technology in AI development serves as a "hedging mechanism" against the monopolies of tech giants. Given that large enterprises may control or monopolize AI technology, cryptocurrencies can ensure that AI remains as open and fair as possible. This is primarily achieved through decentralized AI infrastructure, such as permissionless backend AI computing, storage, data, and model hosting services. In summary, cryptocurrencies will address key issues in a fair and open manner, contributing to the development of the AI ​​field. Part Three: Potential Risks While the combination of Web3 and AI holds great promise, we must also be wary of potential risks. This article will focus on analyzing three major risks: 1. Instruction Injection Attacks Instruction injection attacks are one of the most serious and easily overlooked threats when developing AI agents that can interact with blockchains, wallets, and protocols. An instruction injection attack refers to an attacker manipulating the input of a large language model to make it ignore the original instructions. Refer to the "Diagram of Direct and Indirect Instruction Injection Attacks" for a detailed understanding of the principle. The danger of instruction injection attacks increases significantly in scenarios combining Web3 and AI—because AI not only generates text but also interacts with real assets and protocols. The risk of instruction injection attacks can be mitigated by employing multiple model layers, strengthening system instructions, and other strategies. 2. The Spread of Misinformation With the rapid proliferation of large language models/AI, the risk of misinformation spread is also increasing. In the Web3 domain, this risk can manifest in various forms: for example, using artificial intelligence to forge project announcements or maliciously using large language models to generate false vulnerability reports and audit results. Despite this risk, cryptocurrencies can address this AI-related issue through the approaches mentioned in Part II (such as on-chain signatures) and other methods (such as deploying misinformation detection agents). 3. Mismanagement of Agents When users authorize agents to execute transactions on their behalf, they may face the risk of mismanagement of the agent's funds. When authorized intelligent agents act on behalf of others, issues such as signing malicious transactions, purchasing incorrect tokens, and interacting with high-risk protocols can all potentially occur. In summary, the combination of artificial intelligence and cryptocurrency has a wealth of use cases, and we have only scratched the surface so far.