The economic model of the internet is undergoing changes. As the open network gradually collapses into a “prompt bar,” we must consider: Will AI lead to an open internet, or will it reduce it to a new maze of paywalls? And who will control all of this—large centralized companies, or the broad user community?

This is precisely where cryptographic technology can intervene. We have discussed the intersection of AI and cryptography multiple times; simply put, blockchain is a new way to build internet services and networks that are decentralized, structurally neutral, and can be owned by users. They provide a counterbalance to the increasingly apparent centralization trend in current AI systems, helping to achieve a more open and robust internet by renegotiating the economic relationships behind the systems.

The idea that “cryptography can help build better AI systems, and vice versa” is not new—but it often lacks a clear definition. Some intersecting areas, such as verifying “proof of humanity” against the backdrop of the widespread adoption of inexpensive AI systems, have begun to attract the attention of builders and users. Other use cases may take years or even decades to achieve. Therefore, in this article, we have compiled 11 practical use cases at the intersection of AI and cryptography, aiming to promote in-depth discussions on questions such as “what is possible” and “what challenges still need to be addressed.” These use cases are based on technologies currently being developed, whether they involve processing massive micropayments or ensuring that humans can maintain a relationship with future AI.

identification (IDENTITY)

Persistent data and context in AI interaction

Author: Scott Duke Kominers

Generative AI relies on data-driven approaches, but in many application scenarios, context (that is, the state and background information related to a particular interaction) is equally important, if not more critical.

Ideally, an AI system (whether it’s an agent, LLM interface, or other applications) should be able to remember the type of projects you are working on, your communication style, your preferred programming languages, and many other details. However, in reality, users often need to repeatedly reconstruct this context across different sessions of the same application (for example, when you start a new ChatGPT or Claude session), not to mention when switching between different systems.

Currently, the context between different generative AI applications is basically non-transferable.

With the help of blockchain, AI systems can save key contextual elements as persistent digital assets, loading them each time a session is initiated and allowing for seamless transfer across multiple AI platforms. Moreover, blockchain may be the only technology solution that is both forward-compatible and inherently emphasizes interoperability, which are core attributes of blockchain protocols.

A natural application scenario is games and media involving AI, where user preferences (from game difficulty to key bindings) can remain consistent across different games and environments. However, a more valuable scenario lies in knowledge-based applications, where AI needs to understand the knowledge the user has already mastered and their learning methods; or in more specialized AI usage contexts, such as programming. Although some companies have already developed customized AI robots that can maintain context within a certain range, this context often cannot be transferred between different AI systems within the same company.

Companies have just begun to realize this issue. Currently, the closest to a universal solution is a customized robot with a fixed context. Moreover, within the platform, the practice of sharing context among different users is also beginning to emerge off-chain. For example, the Poe platform allows users to rent their customized robots to others.

By migrating such behavior to the blockchain, we can share a “context layer” composed of all the key elements of our digital behavior with the AI systems we interact with. The AI systems will be able to immediately understand our preferences, better fine-tune and optimize our interaction experience. Conversely, just like the registration of intellectual property on the blockchain, allowing AI to reference the persistent context on the blockchain will also stimulate a new type of market interaction around prompts and information modules. For example, users can directly authorize or monetize their expertise while retaining ownership of the data. Of course, sharing context will also unlock many possibilities we have yet to imagine.

General identification for AI intelligences

Author: Sam Broner

Identification, which refers to the authoritative record of “who” or “what,” is the silent underlying structure that supports today’s digital discovery, aggregation, and payment systems. Since platforms operate these infrastructures hidden behind the scenes, we can only experience its presence in the finished product: Amazon assigns unique identifiers (ASIN or FNSKU) to products, showcases them in a centralized manner, and helps users discover and make payments; Facebook is similar: user identification forms the basis for its content recommendations, Marketplace product displays, and the discovery of organic content and advertisements.

But as AI intelligences evolve, this situation will change. Enterprises are using AI intelligences in various scenarios such as customer service, logistics, and payments, and the platform forms are shifting from a single interface to distributed systems that are cross-platform and cross-terminal. These intelligences will accumulate deep context to complete more tasks for users. If the identification of a certain intelligence is only bound to one platform or market, it will struggle to function in other important scenarios—such as email conversations, Slack channels, or other products.

Therefore, AI agents need a unified, transferable “passport.” Otherwise, we will be unable to identify their payment methods, confirm their versions, query their capabilities, know who they represent, or track their cross-platform reputation. The identification of an agent should include the functionalities of a wallet, an API registry, a changelog, and social proof—so that any interface (whether it’s email, Slack, or another agent) can consistently recognize and interact with it.

There is no unified “identification” primitive, and each integration has to build the underlying structure from scratch, with mechanisms still relying on chance. Users also lose context when switching between different platforms.

We are at a stage where we can “redesign the agent infrastructure from first principles.” So, how do we build a richer, trust-neutral identification layer than DNS records? We should not rebuild the “monolithic platforms” that bundle identification, discovery, aggregation, and payments together; instead, agents should be able to freely receive payments, list capabilities, and coexist in multiple ecosystems without worrying about being locked into a single platform.

This is exactly where the highlights of the combination of AI and cryptocurrency lie: the “permissionless composability” provided by blockchain networks helps developers create more useful agents and better user experiences.

Of course, currently those vertically integrated platforms (like Facebook or Amazon) still provide a better user experience—because one of the complexities of creating high-quality products is ensuring that all modules work together seamlessly from top to bottom. But the cost of this convenience is also high. Especially as the costs of building, aggregating, monetizing, and distributing agents continue to decrease, and the reach of agent applications continues to expand, a trusted and neutral identification layer will grant entrepreneurs a true sovereign “passport” and encourage more exploration and innovation in distribution and design.

Future-compatible “real person identification”

Authors: Jay Drain Jr. and Scott Duke Kominers

As AI becomes more pervasive—whether it’s driving robots and intelligent agents in online interactions or creating deepfakes and manipulating social media—people are finding it increasingly difficult to discern whether their online interaction partners are real individuals or programs. This erosion of trust is not a distant future; it has already quietly arrived. From comment bots on X (formerly Twitter) to bots on dating apps, the boundaries between reality and the virtual are becoming blurred. In such an environment, Proof of Personhood (PoP) is gradually emerging as a key infrastructure.

One of the current ways to verify that a person is human is to use digital identification (such as the centralized identity system used by the TSA in the United States). Digital identification includes various information that users can use to verify their identity—username, PIN code, password, third-party authentication (such as citizenship or credit records), and other credentials. The value of decentralization is evident here: when this data is centrally managed, the identity issuer can revoke access, charge fees, or even assist in surveillance; whereas decentralization reverses this structure, allowing users rather than the platform to control their own identities, making it more secure and less prone to censorship.

Unlike traditional identification systems, decentralized “Proof of Personhood” mechanisms (such as Worldcoin’s World ID system) allow users to self-manage and store their identification data in a privacy-friendly and trust-neutral manner to verify that they are real people. Just like a driver’s license, once a PoP is issued, it can be used across any platform, anytime, anywhere. This blockchain-based PoP thus possesses forwards compatibility, which is reflected in two aspects:

Portability: PoP follows open protocols, and can be integrated into any platform. Identification is controlled by the user and is built on public infrastructure, making it fully portable, allowing any existing or future platform to connect.

Permissionless accessibility: Any platform can independently choose to identify the PoP identification without the need for authorization through a centralized API, thereby avoiding the risk of certain use cases being denied.

The main challenge in this field currently lies in user adoption: although we have not yet seen a large-scale real-world use case for proof of personhood, we believe that once the number of users reaches a critical mass, and with several important partners and certain “killer applications” driving it, the adoption of PoP will accelerate rapidly. Every application that integrates a certain PoP standard will enhance the practical value of that identification, thereby attracting more users to claim that identification, which in turn will encourage more applications to integrate that standard due to the growing user base, creating a rapidly increasing network effect. (And since on-chain identities are designed for interoperability, this effect will be even more explosive.)

We have seen some mainstream consumer applications and services, especially in the fields of gaming, social networking, and dating, announcing collaborations with World ID to help users confirm that they are facing real people, and even the specific real person they expect. At the same time, new identification protocols are constantly emerging, such as the Solana Attestation Service (SAS). Although SAS itself is not a PoP issuer, it allows users to privately associate off-chain data (such as KYC verification or investment qualifications required for compliance) with their Solana wallets, thereby laying the foundation for building a decentralized identity system.

All these signs indicate that the outbreak point of decentralized PoP may be coming soon.

The significance of PoP is not just to “ban bots”; it is a key mechanism that clearly delineates the boundaries between the human network and the AI network. It allows users and applications to distinctly differentiate between “this is interaction between people” and “this is interaction between people and machines”, thereby bringing a safer, more authentic, and healthier experience to the digital world.

Decentralized infrastructure of AI

Decentralized Physical Infrastructure of AI (DePIN)

Author: Guy Wuollet

Although AI is a digital service, its development is increasingly constrained by physical infrastructure. Decentralized Physical Infrastructure Networks (DePIN) as a new model for building and operating real-world systems are helping to democratize the computing infrastructure that AI innovation relies on, making it cheaper, more resilient, and more resistant to censorship.

Why is that? The two main bottlenecks in AI development are energy and chip acquisition. Decentralized energy can help unlock more power resources, while developers are also utilizing DePIN to aggregate idle chip resources from sources such as gaming computers and data centers. These computing devices can collaboratively build a permissionless computing power market, creating a fairer environment for AI product development.

Other use cases include distributed training and fine-tuning of large language models (LLMs), as well as distributed networks for model inference. Decentralized training and inference can not only reduce costs (as it utilizes idle computing power), but also provide censorship resistance, ensuring that developers are not banned for relying on hyperscale cloud service providers.

The concentration of AI models in the hands of a few companies has always been a concerning issue; decentralized networks help build a more cost-effective, censorship-resistant, and scalable AI ecosystem.

Provide infrastructure and safeguards for interactions between AI agents, terminal service providers, and users.

Author: Scott Duke Kominers

As AI tools become increasingly adept at handling complex tasks and multi-layered interaction chains, they will increasingly need to autonomously interact with other AIs rather than rely on human controllers.

For example, an AI agent may need to call certain data related to specific computations or recruit AI agents skilled in specific tasks—such as arranging a statistical robot to perform model simulations or calling an image generation robot during the creation of marketing materials. AI agents will also create significant value for users by executing complete transaction processes or activity flows—such as searching for and booking flights based on user preferences, or discovering and purchasing a new book that matches their taste.

Currently, there is no mature and universal agent-to-agent market—such interactions are mostly limited to explicit API interfaces or a few closed ecosystems that maintain internal agent calls.

A more common issue is that most current AI agents operate in isolated systems, with closed interfaces and a lack of architectural standards. However, blockchain technology can help protocols establish open standards, which is crucial for adoption in the short term. In the long run, this also supports “forward compatibility”: as new AI agents continue to evolve and emerge, they can still connect to the same underlying network. Due to its interoperable, open-source, decentralized, and easily upgradable architecture, blockchain can adapt more rapidly to the changes brought about by AI innovation.

Currently, multiple companies are building blockchain “tracks” for agent interactions: for example, Halliday has launched a protocol that provides a standardized cross-chain architecture for AI workflows and interactions, and has set up guardrails at the protocol level to ensure that AI does not deviate from the user’s intent. Catena, Skyfire, and Nevermind support payment interactions between AI agents without the need for human intervention. Coinbase has also begun to provide infrastructure support for such projects.

Keep AI / vibe-coding applications in sync

Author: Sam Broner and Scott Duke Kominers

In recent years, the explosive development of generative AI has made software building easier than ever before. Coding efficiency has improved by several orders of magnitude, and more importantly—programming can now be done using natural language, allowing even those unfamiliar with programming to fork existing programs and even build entirely new applications from scratch.

However, while AI-assisted programming brings new opportunities, it also introduces a significant amount of “entropy” within and between programs. The so-called “vibe coding” simplifies the complex network of underlying dependencies, but it can also lead to functional or security issues when underlying components are updated. At the same time, as more and more people use AI to create personalized applications and workflows, interactions between different user systems will become increasingly difficult. In fact, even if two vibe-coded programs have the same functionality, their operational logic and output structure may vary greatly.

In the past, the standardized way to ensure consistency and compatibility was through file formats and operating systems, and more recently through shared software libraries and API interfaces. However, in a world where software is continuously evolving, morphing, and forking in real-time, these layers of standardization need to possess broad accessibility and continuous upgrade capabilities—while also maintaining user trust. Moreover, relying solely on AI cannot solve the challenge of incentivizing people to maintain these connections and compatibility.

Blockchain provides a solution that addresses both of these issues simultaneously: embedding a “protocolized synchronization layer” into users’ custom software and ensuring cross-application compatibility through dynamic updates. In the past, a large enterprise might have needed to pay millions of dollars to system integrators (such as Deloitte) to customize their Salesforce system. Today, an engineer might be able to create a visualization interface for sales data over a weekend. However, with the surge of personalized software, developers will also need assistance to keep these applications synchronized and functioning properly.

This is somewhat similar to the operation mechanism of current open-source software libraries, but its updates are real-time, not periodic—and there is an incentive mechanism. All of this can be achieved through crypto. Like other blockchain-based protocols, shared ownership encourages participants to actively engage in improving the protocol. Developers, users (or their AI agents), and other consumers can earn rewards for introducing, using, and improving new features and integrations.

Conversely, shared ownership also invests each user in the overall success of the protocol, creating a “anti-malicious” mechanism. Just as Microsoft would not easily undermine the .docx file format standard because it would affect users and brand reputation, the co-owners of the protocol would also not easily introduce poor or malicious code.

As we have seen with various standardized software architectures in the past, there is also a huge potential for network effects here. As the “Cambrian explosion” of AI programming software continues to advance, the number of heterogeneous systems that need to maintain communication with each other will grow rapidly.

In short: vibe programming must stay in sync, it cannot rely solely on vibe. Crypto is the key.

New economic and incentive model

Support for a micro-payment mechanism with revenue sharing

Author: Liz Harkavy

AI agents and tools like ChatGPT, Claude, and Copilot provide us with a new and convenient way to navigate the digital world. But whether good or bad, these technologies are disrupting the economic system of the open internet. We have already seen preliminary signs of this trend— for example, some educational platforms have seen a significant drop in traffic as students increasingly turn to AI tools; several newspapers in the United States have also sued OpenAI for copyright infringement. If we cannot readjust the incentive mechanisms, the internet will become more closed: more paywalls, fewer content creators.

Of course, policies can also be used to solve problems, but while legal procedures are being advanced, some technological solutions have already begun to emerge. Perhaps the most promising (and technically challenging) is to directly embed a revenue-sharing mechanism into the internet architecture. When an AI-driven action facilitates a transaction, the content creators who provide the information source for this action should receive the corresponding share. This has already been reflected in affiliate marketing systems, where sources can be tracked and revenue can be shared; a more advanced version could automatically track and reward all contributors in the information chain. Blockchain can obviously play an important role in this “source tracing” mechanism.

However, such systems still need to build new infrastructure—especially micropayment systems capable of handling extremely small transactions, attribution protocols that can fairly assess different types of contributions, and governance models that ensure transparency and fairness. Currently, some blockchain-based tools are showing potential, such as rollups, L2 scaling solutions, AI-native financial institution Catena Labs, and financial infrastructure protocol 0xSplits—they can achieve almost zero-cost transactions and more refined revenue splits.

Blockchain can make complex agency payment systems a reality through the following mechanisms:

Nanopayment can automatically split among multiple data providers, allowing a single user interaction to trigger micropayments to all information contributors.

Smart contracts can enable enforceable retroactive payments after transactions, ensuring that information sources contributing to user decisions are compensated after a transaction is completed, while the entire process maintains transparency and traceability.

Blockchain can also achieve complex, programmable revenue distribution rules, enforce profit-sharing schemes through code, avoid centralized subjective judgments, and establish trustless financial relationships between autonomous agents.

As these emerging technologies continue to mature, they are expected to establish a brand new media economic model that covers the complete value chain from content creators to platforms to users.

Blockchain as a registration system for intellectual property and source traceability

Author: Scott Duke Kominers

The rise of generative AI urgently necessitates an efficient, programmable mechanism for registering and tracking intellectual property (IP) — one that can confirm the source of creation while supporting business models around IP for access, sharing, and adaptation. The current IP system relies on high-cost intermediaries and post-facto enforcement, which is no longer applicable in a world where AI can instantaneously consume content and “one-click generate” variants.

We urgently need an open and public registration system that can clearly prove ownership, facilitate efficient operations for IP creators, and allow AI and other web applications to easily integrate. Blockchain is the ideal solution: it allows for IP registration without relying on intermediaries, provides immutable proof of creation, and enables third-party applications to easily recognize, authorize, and interact with these IPs.

Of course, some people are skeptical about the idea of “can technology really protect IP.” After all, Web 1.0 and 2.0, along with the current AI revolution, often come with a weakening of intellectual property protection. But the problem is: many existing IP business models still focus on excluding derivative works rather than incentivizing and monetizing them. Programmable IP infrastructure not only allows creators, brands, and others to clearly establish ownership in the digital world but also gives rise to a new model—building new businesses around a shared mechanism that “allows the legal use of IP in generative AI and other digital applications.”

We have already seen this new attempt in the early NFT field, such as promoting brand network effects and value accumulation through CC0 licensing; further, infrastructure developers have created protocols and even exclusive blockchains (like Story Protocol) specifically for the standardization and composability of IP registration and licensing. Some artists have begun to license their styles and works for creative re-creation through protocols like Alias, Neura, and Titles. The sci-fi series “Emergence” by Incention allows fans to participate in co-creation of characters and worldviews, and retains a record of each creator’s contributions through the Story Protocol registration system.

AI represented by Webcrawler should compensate content creators.

Author: Carra Wu

Currently, the AI agents that best meet market demands are not programming assistants or entertainment tools, but rather Web crawlers – they automatically browse the web, collect data, and autonomously determine which links to visit.

According to estimates, nearly half of the current internet traffic comes from non-human sources. Bots often ignore the robots.txt file (theoretically used to indicate whether crawlers are allowed to scrape the site) and use the scraped data to support the core competitiveness of the world’s largest technology companies. Worse still, websites themselves have to pay for these “uninvited guests,” bearing the costs of bandwidth and server resources. As a result, CDN providers like Cloudflare have had to launch a series of blocking services. Today, this is a fragmented, cumbersome countermeasure, but it can actually be replaced by a more reasonable system.

We have pointed out that the original “economic contract” of the internet—namely the win-win relationship between content creators and platforms—is on the brink of disintegration. This is also reflected in the data: over the past year, an increasing number of websites have begun to actively block AI crawlers. In July 2024, only 9% of the top 10,000 websites were blocking AI scrapers; now, this percentage has risen to 37% and is continuing to rise rapidly.

So, can we stop blindly blocking all requests from suspected bots and instead seek a balance in the middle ground? One new model is: AI crawlers no longer “free ride” on web content but instead pay for data scraping activities. Blockchain can serve as the execution layer of this model: each crawler agent holds cryptocurrency and initiates on-chain negotiation with the website’s “gatekeeper agent” or paywall system via the x402 protocol when accessing the site.

The problem is that robots.txt (also known as “robots exclusion standard”) has become an industry default practice since the 1990s, and overturning it requires large-scale industry coordination or the intervention of CDN providers like Cloudflare. On the other hand, we can create a separate channel for human users: they can continue to access content for free by proving their “human identification” through World ID (see above).

In this way, the behavior of AI collecting content can achieve compensation for creators at the collection points, while human users can still enjoy an internet of “information freedom.”

More privacy-focused advertising: precise yet not intrusive.

Author: Matt Gleason

AI is already changing the way we shop, but can advertisements also be a bit more “useful”? Many people hate advertisements because they can be irrelevant or too intrusive. Even “personalized ads,” if they are too precise and based on a large amount of personal data, can make people feel “spied on.”

Some applications try to monetize through paywalls (such as watching videos or unlocking game levels). Cryptographic technology can help us reshape this logic. Combined with blockchain, personalized AI agents can deliver advertisements based on the preferences set by users without exposing their privacy data; at the same time, they can reward users with cryptocurrency after voluntary interactions.

Technically, this model requires:

Low-cost digital payment system: Advertising interactive rewards must support high-frequency small payments, and the system should have high speed and low-cost characteristics.

Privacy-protecting data verification mechanism: AI advertising agents need to verify whether users meet certain demographic characteristics without exposing specific data; zero-knowledge proof (ZKP) technology can achieve this.

New incentive model: If the advertising revenue model of micro-payments (

Humans have long tried to make advertising more useful, whether online or offline. By reshaping the advertising system to be driven by “AI + blockchain”, it is finally hopeful that advertising can become truly useful: not disruptive and yet profitable.

This will also make the ad space itself more valuable, while potentially overturning today’s highly intrusive “advertising exploitation economy” and instead building a human-centered system: where users are no longer “products”, but “participants”.

Mastering the Future of AI

AI companions owned and controlled by humans

Author: Guy Wuollet

Nowadays, many people spend more time on devices than in face-to-face communication, and this time is increasingly being used to interact with AI models and content curated by AI. In fact, these models have already begun to provide some form of companionship—whether for entertainment, information retrieval, catering to niche interests, or educating children. We can easily imagine that in the near future, AI companions will be widely used in fields such as education, healthcare, legal consulting, and social companionship, becoming a common form of interaction among humans.

The AI companions of the future will have infinite patience and will be highly customizable based on individual needs and specific requirements. They will not just be assistants or “robot servants”; they are likely to become highly valued relational entities for people. Therefore, the question of who will own and control these relationships—whether it will be the users themselves or companies and other intermediaries—becomes crucial. If you have been concerned about content filtering and censorship on social media over the past decade, this issue will become even more complex and personal in the future.

In fact, similar viewpoints have been raised long ago (as seen here and here): blockchain and other censorship-resistant hosting platforms may be the clearest path to achieving censorship-resistant, user-controlled AI. Although individual users can run local models and purchase GPUs on their own, most either cannot afford it or simply do not know how to do it.

Although there is still a distance from the widespread adoption of AI companions, the technology to achieve all of this is advancing rapidly: text-interactive AI companions have already shown excellent performance, and visual avatars have significantly improved; the performance of blockchain is also gradually increasing. In order to make it easier for users to use uncensorable AI companions, we still need to continuously improve the user experience (UX) of cryptographic applications. Fortunately, blockchain wallets like Phantom have made on-chain interactions simple, while embedded wallets, passkeys, and account abstraction technology also allow users to achieve self-custody wallets without having to manage mnemonic phrases themselves.

In addition, high-throughput, trustless computing technologies such as Optimistic and zero-knowledge co-processors will also enable us to establish meaningful and lasting relationships with digital companions.

In the near future, we will shift from discussing “when humanoid digital companions and virtual avatars will appear” to “who has the right to control them and in what ways.”

Statement:

1. This article is reproduced from [BlockBeats] The copyright belongs to the original author [a16z Crypto, Scott Duke Kominers, Sam Broner, Jay Drain, Guy Wuollet, Elizabeth Harkavy, Carra Wu, Matt Gleason] If you have any objections to the reprint, please contact Gate Learn TeamThe team will process it as soon as possible according to the relevant procedures.
2. Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Other language versions of the article are translated by the Gate Learn team, unless otherwise mentioned.GateUnder such circumstances, it is forbidden to copy, disseminate, or plagiarize translated articles.