Ethereum co-founder Vitalik Buterin has called for the integration of transaction simulation mechanisms into crypto wallets and smart contracts to enhance user safety and convenience.
How I think about "security":
— vitalik.eth (@VitalikButerin) February 22, 2026
The goal is to minimize the divergence between the user's intent and the actual behavior of the system.
"User experience" can also be defined in this way. Thus, "user experience" and "security" are thus not separate fields. However, "security"…
"The goal of security is to minimize the gap between what the user wants to do (their intent) and what the system actually does. 'User experience' can be defined in exactly the same way. Therefore, these are not different fields," he wrote.
According to the programmer, security architecture should be built around user intent, creating systems that verify user actions. This applies not only to Ethereum wallets and smart contracts but also to operating systems and hardware.
"The user first indicates what they want to do, and after viewing the simulation of the on-chain consequences, they click 'OK' or 'Cancel'," Buterin explained.
Other options include spending limits and multisignatures. A transaction proceeds only if three factors align: intent, expected outcome, and acceptable risk level. The result should be logical: safe actions are simplified to the point of automation, while risky ones require additional verification.
Implementation Challenges
The key challenge lies in accurately recognizing user intent—a task Buterin described as "extremely complex." This is why there are no "perfect" security solutions.
"[…] Absolute security is impossible. Not because machines or people are 'imperfect', but because the very concept of 'user intent' is fundamentally complex, and the user themselves does not have easy access to it," he noted.
He stated that an effective solution should rely on redundancy: the user expresses their intent in multiple overlapping ways, and the system activates only when these align.
As an additional filter for intentions, LLMs can be utilized, Buterin added. He believes that when applied correctly, neural networks can simulate human logic:
"A typical LLM is (among other things) like a 'shadow' of the concept of human common sense. A model fine-tuned on a specific user becomes a 'shadow' of that user and can more subtly determine what is normal and what is not."
Recall that the Ethereum co-founder discussed the potential process of merging the ecosystem with artificial intelligence.