Over the past period, Vitalik has repeatedly mentioned a somewhat unfamiliar term: CROPS.
The systematic emergence of this concept can be traced back to March 13th. The Ethereum Foundation's Board of Directors released the 'EF Mandate' document, clearly stating that the primary focus will be on Ethereum's censorship resistance, open source, privacy, and security—namely, CROPS—and using this to serve user self-sovereignty while maintaining capture resistance and a more seamless user experience.
This statement is actually very important, especially as AI begins to enter wallet and automated execution scenarios, CROPS is no longer confined to Ethereum's value proposition but could become an issue of whether users can continue to control their digital lives in the AI era.
I. What Exactly is CROPS?
To understand CROPS, one must first break free from a common misconception: Ethereum certainly needs to improve performance and reduce costs, but it's not just competing with other blockchains to see who is faster or has lower fees.
While speed and cost are indeed the most intuitive factors for short-term user experience, if you zoom out and look at the longer term, Ethereum's stance has become increasingly clear over the past two years: what it truly aims to provide is a set of more fundamental capabilities: users can hold assets, express identity, sign transactions, and participate in coordination without relying on a single platform, without surrendering ultimate control, and without being arbitrarily blocked by a centralized service.
This is the significance of CROPS.
In the context of the EF Mandate, CROPS primarily points to five directions, which are also the acronym for their keywords: Censorship Resistance, Capture Resistance (this was actually added later by Vitalik), Open Source, Privacy, Security.
- C - Censorship Resistance: Ensuring transactions and smart contracts are immutable and cannot be halted due to any external political or centralized entity pressure.
- R - Capture Resistance: Preventing Ethereum's governance, development roadmap, and critical entry points from being controlled long-term by a minority of interested parties.
- O - Open Source / Openness: Adhering to completely open-source code and maintaining absolute freedom of access for the ecosystem.
- P - Privacy: Preserving users' right not to be scrutinized through cryptographic technologies on a transparent ledger.
- S - Security: Upholding the bottom line, providing unbreakable ultimate settlement security.
Looking at these items together, they actually constitute a set of selection and guiding principles with a clear orientation, which is also very much in line with Ethereum's consistent value-driven path.
For example, at the protocol layer, it means Ethereum needs to continuously improve censorship resistance, client diversity, validator decentralization, formal verification, etc. At the application layer, wallets, RPCs, browsers, signature interfaces, and account systems also need to reduce dependence on centralized entry points. At the user experience layer, security cannot rely solely on users understanding complex transactions; it must be preemptively integrated through clearer signature displays, more verifiable interactions, and more comprehensive risk warnings before actions occur.
This is also why EF has recently advanced some more specific directions around security, privacy, protocol resilience, and ecosystem public goods, such as the Ethereum Audit Subsidy program, which attempts to lower the barrier for Ethereum ecosystem developers to obtain high-quality security audits. Viewed more broadly, this is not just about subsidizing costs, but about pushing 'security' further from a high-cost service affordable only to a few large projects towards more small and medium-sized developers.
In late May, Vitalik also reiterated his views on EF's future direction, emphasizing that EF should become a smaller, more distinct, and more focused organization dedicated to long-term sustainability, rather than trying to cover every need in the ecosystem. The reason is pragmatic: after all, EF does not possess infinite resources nor a continuous revenue stream from staking or transaction fees. Therefore, it should direct its limited resources towards tasks that are crucial for Ethereum to achieve CROPS values and that other entities find difficult to reliably undertake.
In other words, at this transformative historical stage Ethereum is currently in, CROPS is not an abstract slogan of 'principle over reality,' but rather more like an external definition and constraint on what EF should and should not do.
II. When CROPS Meets AI: The Convergence of Two Parallel Universes
Vitalik Buterin recently brought CROPS into a larger discussion by placing it in the context of AI.
On May 28th, Vitalik Buterin posted an update on his progress with local AI, stating that DeepSeek V4 had released a 2-bit quantized version that could run within about 90 GB of VRAM, achieving speeds of around 35 tokens/second on Apple hardware and about 7 tokens/second on AMD hardware. He noted that a true 'CROPS AI' should support multiple hardware platforms, not just be 'decentralized AI.'
Simultaneously, he pointed out significant overlap between the CROPS Ethereum access layer and CROPS AI, for example, using zero-knowledge proofs for paid remote LLM calls and private Ethereum RPC reads. He suggested there should also be more AI models fine-tuned for Ethereum scenarios in the future, aimed at improving smart contracts, protocol code, and ecosystem security.
This essentially places Ethereum and AI within the same problem framework.
Previously, discussions about AI often focused on model capabilities—whether they can write code, and especially whether they can perform complex tasks as human substitutes. However, from a user safety perspective, the real change brought by AI is not just 'increased capability,' but that it is altering the entry points for digital operations.
To reiterate an old point: in the past, applications were relatively clear, separate interfaces. We opened wallets to transfer funds, opened DApps to trade, opened browsers to search, opened social products to post. Each application had relatively clear boundaries. But with the emergence of AI Agents, these boundaries are becoming increasingly blurred. Users no longer click functions one by one; instead, they express intent in natural language:
Help me find the optimal cross-chain route, help me make a swap, help me organize my assets, help me execute a certain DeFi strategy, help me generate and send a transaction...
This sounds convenient, but it also implies a more critical question: when AI becomes your digital agent, what transactions is it signing on your behalf, and what privacy is it exposing?
If AI runs entirely on centralized cloud services, users' asset information, transaction intent, address relationships, identity preferences, and operational habits could be concentrated in the hands of a few service providers. Especially when executing on-chain operations relying on opaque APIs, centralized RPCs, black-box plugins, and unverifiable inference processes, users might become more convenient but also less aware of what they are actually surrendering.
This is the question CROPS AI must answer.
A CROPS-compliant AI is not just about being capable; it should also be as censorship-resistant, open, privacy-preserving, and secure as possible. It should ideally run locally, or at least minimize reliance on centralized cloud services for sensitive scenarios, minimize information leakage, and allow users to understand, confirm, and retain ultimate control.
In other words, AI cannot just be a smarter black box. Especially in Web3 scenarios, AI in the future may not only help you summarize articles, write code, or provide customer service, but directly participate in asset management and automated execution.
The closer it gets to user assets, the more important CROPS becomes.
This is also why there is overlap between the CROPS Ethereum access layer and CROPS AI.
III. What Web3 Incremental Opportunities Lie in This Intersection?
From this perspective, Vitalik's recent mention of the overlap between CROPS Ethereum Access Layer and CROPS AI is very natural.
Because whether it's Ethereum or AI, the core problem users face is becoming the same: how can I use AI assistance without completely handing over my privacy, identity, assets, and right to choose to centralized intermediaries?
- On the Ethereum side, this manifests as: How do users access on-chain data? How do they connect to RPCs? How do they sign transactions? How do they confirm if a DApp interaction is safe? How do they avoid having all wallet queries, balance reads, and transaction broadcasts pass through a few centralized services?
- On the AI side, this manifests as: How do users call models? How can they ensure prompts and personal data are not misused? How can local models handle sensitive tasks? How, when needing remote large model capabilities, can they minimize exposure of their identity and intent?
These two sets of problems seem different, but are fundamentally similar.
For example, when Ethereum users query balances, read transaction history, or simulate transaction outcomes, they often need to go through RPC services. While RPCs appear to be mere technical interfaces, they can know your IP, addresses, query habits, asset structure, and interaction paths. If such data is centrally collected, users' on-chain privacy can be gradually pieced together.
But AI users calling remote models may also expose their preferences, financial information, or even identity clues. If users employ AI for wallet operations in the future, the risks would be further amplified.
Therefore, the ZK-paid remote LLM calls and private Ethereum RPC reads Vitalik mentioned essentially attempt to solve the same problem: how can we obtain services while invoking remote capabilities without exposing all our information?
This is where CROPS Ethereum and CROPS AI intersect. On one side is a more private, verifiable, and lower-trust-assumption on-chain access layer; on the other is a more open, localized, and secure AI execution environment. Combined, they may form a new entry point for users into the digital world in the future.
Extending outward along the underlying logic of CROPS, the entire Web3 ecosystem (especially the wallet layer as a traffic entry point) will undoubtedly take on more roles.
As users begin to express on-chain needs in natural language, a wallet is no longer just a signing tool but the control console for a user's digital actions. It needs to help users determine: Can this DApp connect? What exactly will this transaction do? Is this AI Agent calling unnecessary data?
From this perspective, CROPS is not an abstract value but will directly influence the design direction of wallet products and drive the development shift for the next decade in integrated Web3 interaction experiences and the wallet sector.
Conclusion
Although, in the current market sentiment, many people might not pay as much attention to purely conceptual topics anymore.
But the colder the market, the easier it is to overlook those technical variables that may not be 'sexy' in the short term but truly determine the direction in the long run.
CROPS is worth paying attention to, not because it creates a new hype cycle, but because it reframes the long-term problems of Ethereum and AI within a single framework: as digital systems become increasingly powerful, can users continue to retain their own control?
After all, security and privacy cannot just be patches applied after the fact.
From this perspective, in an era where AI is accelerating its takeover of the digital world, these factors may be the truly positive variables that make Ethereum worth continued building and usage.
In an era where AI is accelerating its takeover of the digital world, being more understandable, more verifiable, more private, and more secure may be the real reasons Ethereum continues to be worth building and using.
