A transaction-filtering system that security startup Phylax rolled out with block builder Titan Builder is drawing fire from Ethereum researchers and a digital-assets lawyer, who say the tool could harden the network's already lopsided block-building market and, in some configurations, leave smart contracts effectively frozen.
Phylax unveiled the system on Nov. 10. It lets decentralized finance applications write rules that screen out transactions capable of producing harmful outcomes while a block is being assembled, rather than waiting for monitoring tools or emergency intervention after an exploit has begun. Titan, which according to Phylax produces one of every two Ethereum blocks, checks pending transactions against those rules before placing them in blocks it builds.
The company frames the mechanism as a circuit breaker. In its announcement, it describes a hypothetical vulnerable vault that could mistakenly release $1 million to a user whose shares are worth $1,000, and says the system could cap how much leaves the vault in a single transaction. Phylax's stated goal is to keep hacks and logic errors from ever settling. "Our mission: stop users from losing money to hacks and errors in DeFi," the company said in a blog post.
Phylax founder and chief executive Odysseas Lamtzidis said on X that DeFi platforms Lighter and Derive are the first customers, with almost $1 billion in assets protected between them.
The debate turns on what happens when the checks are not available. Digital-assets lawyer Gabriel Shapiro warned that contracts using the system can be written to reject transactions unless they arrive in blocks from builders that have performed the required attestations. If those builders stop attesting, the protected functions could keep refusing transactions until checks resume, giving applications a de facto freeze switch.
Lamtzidis said protocols can avoid that outcome by adopting what he called a fail-open setting, which lets transactions through when the security checks cannot be verified. "The sane default we provide and suggest to protocols is a fail-open behavior," he said. The choice remains optional, however, and he noted that some protocols may decide to put safety ahead of liveness. Operators can also choose whether their contracts are immutable or changeable through governance.
Shapiro argued that fail-open behavior should be mandatory rather than voluntary, and that any change to the safeguards should be subject to a time delay so that protections cannot be stripped at short notice.
Ethereum researcher Fan Zhang drew the line between filtering that applications choose and filtering imposed from outside. "Applications willingly signing up for filtering is fine, as a user can walk away from such applications," he said. "It's externally forced censorship that's the problem. That's where I'll draw the line."
Stefanie Boss, a University of Amsterdam researcher who co-authored work on censorship resistance and compliance behavior in Ethereum's consensus layer, said rules aimed at what a transaction does are less troubling than rules aimed at who sends it. She called the result-driven design comparatively low-risk and the security objective legitimate, arguing that the damage from major hacks far exceeds the harm of this kind of exclusion. Even so, she cautioned that similar rules could be turned toward sanctions or politically motivated exclusions. "It is important to ensure that there are proper safeguards in place to make sure that these rules will remain fair and designed for security," she said.
Concentration cuts both ways
Zhang said the approach only became workable because of how concentrated Ethereum's block-building market has grown. "We finally have a way to implement this effectively, but sadly only because of how centralized Ethereum's builder market has become," he said. He sees real security value, particularly as AI agents hunt for bugs faster than human reviewers can, but warned that inserting extra code at such a sensitive stage of transaction processing opens new failure paths, including implementation mistakes and the theft, loss or impersonation of builder signing keys.
Sen Yang, another Ethereum researcher, looked at the competitive angle. If major protocols come to depend on a handful of builders to enforce security, those builders could gain an edge that deepens concentration. He said the proposed enshrined proposer-builder separation upgrade, expected in the Glamsterdam release, may reduce reliance on third-party relays but does not directly tackle builder concentration. Transactions that dominant builders exclude could still be picked up by builders that do not filter, he added, which would mean delays for users rather than permanent exclusion.
The dispute lands as Ethereum developers work through related builder-key issues. Developers merged a Glamsterdam fix for a builder key flaw in October, and a Cambridge-led study in July flagged node concentration as a risk to the network.
What to watch
The practical stakes for DeFi users hinge on configuration choices that are still being made. Protocols that select fail-closed behavior would tie access to protected contract functions to the availability of builder attestations, turning an external dependency into an outage risk during upgrades or shifts in builder policy.
Note: Details drawn from Phylax's announcement and interviews with researchers and a lawyer published around it.
Phylax rejects the censorship label. Lamtzidis said the system is explicitly opt-in because applications themselves write the rules, and compared the checks to conditions already hard-coded into contracts. "It's no different than a 'require()' statement," he said. "Censorship implies arbitrary decision-making or coercion."
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.



