Crypto Assets and AI: It's Not Just Hype, But a Major Transformation for the Future!

Compiled: Vernacular Blockchain

AI is the area I am most excited about in the cryptocurrency space. But most people think AI is just a trendy buzzword in the web3 domain.

Today, I want to unveil this layer and analyze the true intersection that can change this field and has immense potential. I believe that crypto x AI will change the way the crypto market operates through practical use cases, while addressing the main issues faced by AI. I have divided this article into three key parts:

• Part One: AI x Web3 ( Practical Use Cases )
• Part Two: How Web3 Facilitates AI Development
• Part Three: Potential Risks

Part 1: AI x Web3 (Practical Use Cases)

For ordinary people, this part may be the most important, as it delves into how you can actually utilize AI in Web3.

1. Transaction Management

One of the most interesting tracks for applying AI in Web3 is using trained agents for trade management/execution. For example, future trades will no longer be executed manually but will instead deploy a “personal assistant” (AI agent) that can trade, execute, and manage positions on your behalf.

The Future of Web3 Proxies These proxies can not only execute trades and manage portfolios for you (we have already seen some protocols being built), but they will also be able to interact completely on-chain. @HeyAnonai is a great example. With their AI protocol, you can already trade, perform cross-chain operations, and interact on-chain using natural language prompts.

! [image] ()

2. LLM (Large Language Model) in Web3

Most people think that using LLM in Web3 is just about asking questions to ChatGPT, but its significance goes far beyond that. You can think of LLM as an interface layer between humans and protocols—this means they can access data through natural language.

LLM for Web3 In practice, this means that as LLMs enter a stage of large-scale adoption, they will also eliminate a significant barrier to accessing market information, as more protocols will be built specifically for training LLMs on Web3 data. Imagine this: if you could access top data/information 24/7 with simple prompts, how much more successful would your trades be?

! [image] ()

3. Security/Privacy

AI can trigger lightning alerts in seconds, which is much faster than most people. These alerts are triggered when AI models analyze on-chain transaction sequences to detect typical exploit behaviors. The significant advantage of AI models in this field lies in their pattern recognition capabilities. For the average person, this means fewer exploits and safer smart contract interactions.

Part Two: How Web3 Facilitates AI Development

In this section, I would like to delve into a part of @a16z's latest report on the “2025 Cryptocurrency Status”.

! [image] ()

In this report, they highlighted four specific AI problems that Web3/cryptocurrency can address:

1. Verify human/AI activities
2. AI agents participate in economic activities
3. Promote appropriate IP licensing
4. Maintain the fairness and openness of AI

Although solving these AI problems is not easy, cryptocurrencies have proven to have some of the best tools. Let's break it down one by one:

1. Verify Human/AI Activities

AI is still considered to be in the early stages of development (relative to its future), but the issue of distinguishing between human and AI activities has become very serious. Cryptocurrency can address this issue in three main ways:

• PoH (Proof of Humanity) System: Verifies that there are real people behind operations without revealing identity (doxxing). Protocols like Worldcoin are already looking to apply this approach on a large scale.
• On-chain signature activities: When the AI agent acts on behalf of the user (as described in the first part), these actions can be signed through cryptography.
• On-chain Accountability: Every action taken by AI agents (whether executing transactions, voting in a DAO, or modifying data) can be recorded, verified, and traced on the blockchain.

2. AI agents participating in economic activities

As emphasized by a16z, to fully unleash the power of AI agents, they need to be able to participate in economic activities. This is where x402 (and other tools) come into play.

! [image] ()

@suhailkakar Simply put, it's like adding a wallet to the internet. It’s a new web standard that allows websites to say: “Hey, before I give you this data, please make a small payment.”

x402 is just an example of how cryptocurrency enables agents to engage in economic interactions. In addition to Suhail's infographic, I also created my own x402 cheat sheet:

! [image] ()

x402 - An explanation that a five-year-old can understand

3. Appropriate IP License

Ensuring that owners have appropriate intellectual property (IP) rights is crucial for the fair use of AI technology. By putting these rights on the Blockchain, we can verify their legitimate use, and the licensing terms can be embedded in smart contracts. In this field, one protocol I like is @campnetworkxyz. Their protocol allows anyone to own their IP and monetize it.

The concept of Camp

! [image] ()

4. Maintain the fairness and openness of AI

Finally, using Web3 technology in AI development is a way to “hedge” against Big Tech. Since large companies are likely to own/monopolize AI technology, cryptocurrency ensures that we can keep AI as open and fair as possible. This is primarily achieved through decentralized AI infrastructure. That is, permissionless backend AI computing, storage, data, and model hosting.

Summary: Cryptocurrencies will address major issues fairly and openly, helping to promote the development of the AI field.

Part Three: Potential Risks

Although the future of Web3 x AI undoubtedly has immense potential, we also need to be aware of the potential risks. In this article, I will focus on three of these potential risks:

1. Prompt Injections

As we build AI agents that interact with blockchain, wallets, and protocols, one of the biggest and least understood threats is prompt injection. You can think of prompt injection as an attacker manipulating the input of the LLM to make it ignore the original instructions.

! [image] ()

Direct vs. Indirect Prompt Injection (Illustration)

In the context of Web3 x AI, prompt injection has become more dangerous because AI is not only generating text but also interacting with real assets and protocols. The risks of prompt injection can be mitigated by employing multi-layer models, enhancing system prompts, and utilizing several other strategies.

2. Intensifying the Spread of Misinformation

As the use of LLM/AI grows rapidly, the risk of spreading false information is also increasing. This can manifest in various ways in Web3, from using AI to forge project announcements to the malicious use of LLM to create false vulnerability reports and audits. While this is a risk, cryptocurrency can also help address this AI issue through the methods discussed in the second part (on-chain signatures) and other approaches, such as deploying misinformation detection agents.

3. Improper Management of Agents

When users authorize agents to execute transactions on their behalf, there is a risk of mismanagement by the agent. For example, signing malicious transactions, purchasing the wrong Token, and interacting with risky protocols are all real risks when enabling agents to act on our behalf.

This article link:

Source: