Balancer hack: hard fork

General
41

You’re arguing in bad faith.

In both cases, no one forced anyone to do anything.

If the hack had been a bug that sent all the funds to 0x000000, you would have made the exact same decision; the hacker’s malicious intent has nothing to do with it.

The only difference is that in one case, Gnosis LTD chose to act by doing a hardfork, and in the other, they chose not to protect their users by keeping the vulnerability to themselves, rather than warning their communities about it.

I never talked about doing a hardfork for EURe/sDAi; from a personal standpoint, I was actually against reimbursing the LPs. You’re using a straw man argument.

42

I agree with the action taken by the Gnosis team to block and recover the stolen funds. Given the circumstances, it was likely the best possible option. However, at the same time, it operated at the edge of the principles of decentralization and immutability.

This was not the result of a single bad decision, but rather the consequence of missing predefined rules or a clear “what-if” framework for handling such incidents. This clearly highlights the need to define such rules going forward—and the sooner this framework exists, the better—especially if Gnosis aims to position itself as a chain with a strong reputation for secure and reliable decentralized finance.

From my perspective, an ideal approach would be to give validators the ability to immediately block suspicious addresses at the validator level, for example via a sanction / block list. This would allow for a fast response and prevent attackers from moving funds without requiring emergency client modifications or rebuilds.

Any subsequent action—such as fund recovery, unfreezing, or other resolution—should then be governed by a predefined framework or by validator governance (e.g. voting), rather than being decided ad hoc.

To support this, it would be beneficial to integrate functionality directly into execution clients that enables validators to manage sanctions lists—manually or automatically—and to keep funds frozen until the appropriate framework-driven or governance-driven process is completed.

1 Like
43

To me this is the only good reason I’ve heard so far to rush this hard fork and make it happen before a proper framework is in place. If there’s a real risk of the recovery failing if this hard fork is delayed by a few weeks, then it may indeed be worth doing it as soon as we can, with a commitment to also get the framework done as soon as possible.

There is no way to force validators to follow them but it does make their decision much easier if they can defer to a framework that has already been agreed upon by the broader Gnosis Chain community (instead of having to decide by themselves which protocol and victims are worth helping out should this ever happen again).

You’re not missing anything, I was just combining the three sorts of feedback ( unequal treatment / precedence / future situations ) that were brought up across this thread and the two previous Balancer-hack-related threads ([1] [2]) . Unequal treatment was brought up in connection with the Balancer oracle pool situation in the past, but this kind of discussion is guaranteed to come up again in the future if there is no framework.

2 Likes
44

I want to voice my support for the hard fork proposal, though I share several of the concerns raised by others in this thread.

Why I Support the Fork:
Ultimately, I agree with @filoozom and others that the “immutability” argument was effectively nullified when we implemented the soft fork. We have already intervened to freeze the funds; refusing to take the final step to return them doesn’t preserve our neutrality and it just leaves victims without recourse while the funds sit uselessly onchain.

If Gnosis Chain is positioning itself as a pragmatic, “Neofinance” layer, we have to prioritize user protection over dogmatic adherence to “Code is Law,” especially when the thief has already been stopped. Leaving $9.4M permanently locked in a hacker’s wallet helps no one and only hurts the reputation of the chain as a safe place to do business.

That being said, the valid concerns raised by others regarding precedent cannot be ignored.

  • Where is the line? If we fork for $9M, do we fork for $1M? Do we fork for individual user errors?
  • Moral Hazard: We must ensure that application developers do not start viewing the Gnosis validators as an insurance policy, leading to lower security standards in their own contracts.

So again, I support this hard fork because it is the right thing to do in this specific instance where funds are already frozen. However, I believe this action must be coupled with a commitment to establish a clear framework for future interventions. We need defined thresholds (e.g., % of TVL affected) and strict criteria so that future decisions are based on established rules, not community sentiment or political pressure.

3 Likes
45

I want to respond specifically to the repeated comparison between the sDAI–EURe pool incident and the Balancer hack, because I think this is where we are talking past each other.

First and foremost, I agree that we made a mistake in how the issue was communicated when it was first encountered. I was involved in and supported that decision at the time, and I’ve explained the reasoning before: this pool was a core piece of infrastructure and an enabler of Gnosis Pay. After years of work, it had finally reached meaningful liquidity, and the intention was to first fully understand and fix the issue before going public, with the goal of making LPs whole. With hindsight, that tradeoff was not handled well.

When you reached out to me privately, @NolanV, I explained this, and later also posted it on the forum. We disagreed on whether the fix sufficiently reduced LP risk, and we didn’t reach consensus, which is fine. It’s also worth noting that this was less than a week after we had asked Balancer to increase the pool fees, and the situation was still being actively monitored.

The subsequent delays in public communication were a process failure involving multiple parties. karpatkey had taken responsibility for communication and repeatedly postponed it. When it became clear this was not progressing, the issue was ultimately made public, and the DAO has since addressed the breakdown in that process.

There are many valid criticisms one can make about how this was handled. What I don’t see is how this supports the claim that the Balancer hard fork decision was driven by favoring certain LPs.

If the underlying suggestion is that we reacted differently in the Balancer case in order to favor LPs, that explanation does not hold. A substantial share of the Balancer exploit directly involved the sDAI–EURe pool, meaning there was significant overlap in the affected LPs. The same LPs are therefore being invoked to explain two opposite, incentive-driven behaviors. The difference in response cannot be explained by who the LPs were, but by the nature of the incidents and the response options that were actually available.

4 Likes
46

That’s the very heart of the matter.

In the case of the Balancer hack, you took the initiative to protect your community without waiting for a third party to intervene. Which is understandable, I don’t have a firm opinion on the matter.

However, in the case of the oracle vulnerability, you waited for months for someone else to take care of the problem.

And I also want to point out that the problem was still not fixed; the oracle was still vulnerable, causing LPs to lose money, as well as GnosisPay users who were unlucky enough to make a swap over the weekend.

That’s also not true, because the original LPs in the EURe/sDAI pool didn’t lose that much money. They certainly weren’t compensated for the risk they took, but they didn’t lose all that much.
The biggest losers are those who joined the pool later(and/leave before the cache fix) and therefore suffered the majority of the losses. To confirm this, you can check the historical price of the pool’s BPT.

This isn’t an opinion; Gnosis LTD reacted differently to two similar situations.

In both cases, your communities deposited money into Balancer and lost it due to an “unknown-unknown.” Whether it was a problem in the bytecode or in the storage is just a detail; there was an attack, and there were losses.

In the first case, the moral solution was to warn the LPs and ensure the pool was deprecated on the Balancer front-end. In the second, it’s to do a hard fork and return the money to its rightful owners.

In both of these cases, Gnosis had no contractual obligation to act—it wasn’t an execution or consensus bug in a client. You made a choice, and a choice must be defended with consistency.

47

If you indeed don’t have a firm opinion on this matter, let’s continue this discussion in a topic which is specific to the issue you’re pointing out: the sDAI-EURe Balancer pool.
In any case, I already recognised that we made a mistake and should have communicated earlier. What else do you expect to get from this?

You also recognise that this is the moral thing to do.

There are a lot of very important issues raised by the community on this topic, mainly the lack of a proper framework to determine when it’s correct to act, for reference of future cases. I think we’ll all be better off if we focus the discussion on that.

1 Like
48

Hey guys.

Thank you for sharing your views.

A few things I would like to address:

  1. I said that there would be a DAO vote on doing a hard fork for the recovered funds and then we went to prepare the binaries for a hard fork without that vote. This is on me. I simply forgot that I said it, I wrote the blog post after a few very short nights. You are right that until now, we haven’t done anything irrevocable, if validators were to choose to run the pre soft fork clients again, the state would be exactly as it was before and the hacker could send transactions moving that balance that get included in the chain.
    I still think doing the hard fork now without a vote is the right thing to do for several reasons:
    → The validator set is already going to cast a vote by upgrading or not upgrading their node.
    → If we run a vote now, results will come in five days, one day ahead of the planned hardfork. Some nodes already upgraded, with one day after results come in two days before Christmas, this will be a total mess and we risk network instability.
    → Additionally, our chances of successful recovery diminish gradually, I don’t want to delay all of this into the new year.
    → The Chain is in a messy state with gas and transaction finalization issues owing to the soft fork, I would be loath to drag this out until the new year. We can also only do Fusaka after this, because it will remove all censoring logic from the clients.
    → We will have a discussion and a vote what to do with the recovered funds after Christmas.

  2. Framework when to attempt a recovery. Thank you for thoughtful comments (@MichaelRealT , @Ignas etc). Parameters that I would like to see included in this framework:
    → funds affected as a percentage of Chain TVL
    → nature of the protocol: if you’re affected by the third bzx hack, you deserve less consideration by the Chain than if you had your funds in a blue chip that’s been around for 5+years and has been audited by everyone and their grandma (like here).
    → This goes even more so if normies are affected. Our ability to gauge defi risks is much more advanced than the average normie lured in by low risk defi promises.
    → technical feasibility: how complex is the recovery in the execution? I.e. had the balancer hacker started moving funds from address to address, would we have attempted a soft fork with a dynamically updatable censor-list?
    I would suggest scoring percentage of chain TVL and nature of the protocol hacked on a score from 1-5, possibly add one or two bonus points for “normie” users (questionable, please weigh in!) and then compare this with the riskiness of recovery.
    So >1% TVL → 1 point, >3% → 2 points, >6% → 3 points, >10% → 4 points, >20% → 5 points
    Bluest blue chip affected → 5 points, newer/ more complex/ more risky protocols then 4, 3, 2, 1 points.
    Are there other factors you would like to see reflected?

  3. Credible neutrality: In order for a base protocol to be credibly neutral, it needs to be blind. And not wilfully blind (I will not look at transactions that I process) but actually blind (I cannot look at transactions that I process). Look at this panel I did with Peter van Valkenburgh at devconnect – relevant part starts a 6’30. Peter says “ultimately transparency will destroy neutrality” and reasons that individual validators will in the mid to long term be prosecuted for including nefarious transactions in blocks, and the only way to bypass this is by making it such that the validators cannot know what they are processing. This is something that we need to double down on. We started this a while back with shutter, but at the time there wasn’t demand. Maybe this is something that we have to push on the ecosystem? What are your thoughts here? Anyways – I would argue that while we can still do something, under certain conditions (see point 2 above) we have a moral obligation to do so.

9 Likes
49

So much vocal activity here. This forum post itself must have contributed to a significant portion of the internet’s non AI generated content over the period that it has been active. Good work all remembering how to write unassisted by LLMs :laughing:

In all seriousness, this is a validator question, and they are presented with software by which to upgrade to, or not. DAOs have no vote on this - that is simply not the way the consensus works at the network level. Even if the DAO did a vote, it is NOT enforceable.

Votes are already baked into the network - except it’s on the Gnosis Beacon Chain, not some snapshot. You want to vote on it? Run a Node | Gnosis Chain - anything else is just theatre.

8 Likes
50

No it’s not. More accurately it is called “communication”.

Generally agree on your take - in the systems’ consesnsus only thing that matters is what the validators choose to update.

However it is not detached from the “community” or the DAO - the DAO is the steward, isn’t it?@serenita_luca as any responsible validator would be doing is checking on community sentiment.
The “theater” is reflecting community sentiment and offering ways for positive engagement + understanding (finding some critical blindspots).

51

My comments are limited to the act of voting, in that there is only one non theatrical method to vote. That is on the beacon chain.

52

My criticism was the implication that the “theater” aka communication is unnecessary.

2 Likes
53

Very important to go there IMO - while it would hamper the capability of such “fixes” it is ultimately necessary to not make the validators responsible/accountable.
The validators should be in the spot of simply ensuring the execution of the code and not worry about “political” decisions.

I would argue that the visible commitment and progress to go there can balance out the ugly taste of the current situation.

1 Like
54

I’m speaking on behalf of myself and not fully for the Lodestar team or ChainSafe as there are varying opinions on this topic.

I’ve always valued Gnosis Chain and their credibility for upholding the values of decentralization and neutrality, arguably more so than Ethereum. I’d like to also state that we’ve had a good relationship with Gnosis, ensuring that Lodestar was always an option for Gnosis validators and recently expanded that support by running some validators for the network. I haven’t been super active in following and participating in the governance of this chain, so I was oblivious to this entire situation until recently - which admittedly, is partially my fault. In fact, I was so oblivious to this situation, that I had no idea validators were distributed custom images to run for actively censoring the attacker on the network. We were still running non-censoring images with our Gnosis validators as of yesterday.

Needless to say, I’m very disappointed in this situation and how the response was handled - not only of the censoring validators, but also the client teams that provided custom, censoring images to a non-consensus breaking bug. We were also left out of the backroom distributions of custom images + conversations, similarly to solo stakers on what was happening in response to this exploit… which shouldn’t be happening anyway on a “decentralized” network. We can debate moral interventions and what the “right thing to do” is all day, but the point I’m trying to make is that this has no doubt destroyed the credible neutrality that I once believed was part of this chain’s ethos. I believe that only time and our future actions from here can repair most of that damage. Similarly to Ethereum’s DAO hack - it’s impossible to fully regain that trust, but it is possible to regain most of it… and it won’t be an easy feat. Additionally, it naturally becomes harder to intervene as more value gets built on top of a chain. For example, this type of intervention would not work very well for Ethereum today.

Now onto solutions: We intend to continue validating and will participate in the hard fork as the damage to credible neutrality was already done when the majority of validators decided to start censoring. We can at least turn a lose-lose situation into lose-win for the victims. We will continue supporting users of Lodestar on Gnosis as we do not want to abandon them in any way for something they did not do - we want to continue empowering solo stakers.

I believe having an intervention framework would be a transparent thing to do and useful in the future as guidelines because I don’t believe this hack was significant enough to have justified this type of intervention in the first place. To use Ethereum again as an example, the DAO attacker had 14% of ETH at the time of the hack. Gnosis had about $9.4M worth of assets exploited over about ~$230 150M of TVL. The tradeoff of credible neutrality for recovering ~ 4 6% of assets was not worth it IMO. But what’s done is done. Let’s just focus on what we can do going forward , rather than what has already happened.

3 Likes
55

My request was very simple: I wanted to know why Gnosis LTD handled these two situations differently. But after reading your response carefully, I think I understand that you consider this to have been an error on your part and that if a similar situation were to arise again, you would act more promptly. Is that correct?

If that is indeed the case, I have no further objections in this discussion.

Before creating a framework, it was important to settle the issues I raised.

  • You’ve addressed the differential treatment.

  • We just got the answer for not consulting the DAO. A little “oopsie, I forgot the vote, haha I’m so distracted :face_with_hand_over_mouth:”—a perfectly valid reason. I myself forgot to take out the trash yesterday, so what’s a vote… just a minor detail.

  • I’m still waiting for the list of wallets affected by the hardfork, as well as the amounts, because yes, this is important information—transparency, transparency, transparency.

  • Finally, it would be interesting to know how many validators are run by the directors of Gnosis LTD, as well as by Gnosis LTD itself. Cf. my previous message about the concern that the network’s decentralization could be artificially inflated.

This is what I’m expect to get from this.

Of course, nothing in my requests is a blocker for the hardfork, but it will be interesting for the next chapter of Gnosis’s fabulous adventure.

1 Like
56

Hi everyone, Danko from Balancer here.

We have been closely following this discussion and want to congratulate the Gnosis community for the thoughtfulness and rigor you all have brought to this debate.

I fully understand and respect the arguments regarding credible neutrality and immutability. The philosophy that a blockchain should remain agnostic to the transactions it processes is a cornerstone of this industry. Everyone recognizes Gnosis Chain operates as neutral infrastructure, however I don’t agree that intervening for an application-layer exploit could set “unwanted” precedent.

At its core, this proposal was driven by a desire for restitution and security; the intent to make users whole and refuse to accept theft as a “status quo”. At Balancer, we deeply appreciate the work being put to protect liquidity providers who have been harmed.

Either path the Gnosis community chooses to take, we value the shared desire to protect users. We remain committed to working with Gnosis, whitehats, security partners and law enforcement to explore every other avenue for recovery.

10 Likes
57

Thank you for engaging in a robust discussion about important issues for the DAO - comms, transparency and operations. The Gnosis team hears you and wants to continue to hear from you.

The next community AMA with the Gnosis core team is on Wednesday, January, 7th 2026. Please share your questions by following this link:

2 Likes
58

i brought quite a crowd of people to gnosischain/gnosispay and told them to support the ecosystem by LPing in EURE/SDAI and use their funds via gnosispay card.
now everyone is thefted of their funds.
imagine what these people call me now.

59

Lots of discussion here after my last intervention, so I couldn’t comment or respond to all the messages.

First of all, thank you to the people who have responsibility in the Chain and protocol who intervened, I strongly regret that all these exchanges did not take place further upstream as had been promised to us, probably one more oversight just like for the DAO vote…

Reading the discussion thread several times to try to understand everyone’s opinions, I could conclude that there are 2 parties trying to make their point:
the first the actors who decided and worked on the fork accompanied by the users who lost funds in this hack story, I will name them in the rest of the answer the Fork-approve
The second group, which mainly defends the transparency, democracy, and original value of the blockchain, is called the Fork-reflections.

My comments here will try to be as objective as possible but it is possible that this is distorted by my membership in the Fork-reflexions group.

The main problem that seems to stand out from the others in the discussion thread is not the fact of having to do a fork or not, the major problem here is the approach that was used for the announcement and the steps to achieve it.
This causes a loss of confidence in the actors who undoubtedly have significant power over the chain and therefore the first value defended by the Fork-reflexions group.
Decentralization does not simply mean having thousands of nodes, it is measured by the capacity of an entity or a group to put pressure on or influence the choice that the validators will make.

I am not saying that Gnosis LTD or any other affiliated entity has bad intentions, that would not be in their interests, what I am saying is that if an actor (Gnosis LTD or any other or restricted group of actors) can influence a decision, then it is not decentralized, and this poses a real problem for the chain.
The problem is not that Gnosis LTD has bad intentions, but the company could be subject to government persions that would go against the interests of users.

Example, in Europe (which is no longer a democracy for several reasons that I will not describe here) the unelected leaders decided to want to read all the messages from everyone in a continuous flow, create databases and profiles, they also decided that all the participants who bridge the gap between crypto and FIAT had to provide in a continuous flow all the detailed information on the transactions, they will store this in a Database.
In France a few days ago the Ministry of the Interior was made private, it seems that a significant amount of highly sensitive data on the population has been stolen, this creates a real individual security problem.
I give this example to explain that the pressure on an identifiable entity is not a fantasy. The question is not whether it will happen, but when it will happen.
When this happens, we could very well see the EU putting pressure on Gnosis LTD or another actor to gradually push updates that are against users.
We can think that we have nothing hidden and that it does not pose a problem to the person who has done nothing wrong, but it is a matter of course what can happen, if the EU has debt problems and can no longer pay this debt, the creators will request the seizure of the property of the inhabitants of the EU, it would therefore be possible with a fork approach which can be executed or influenced by a group or an entity.
In this scenario (also in others) even a person who has done nothing wrong could see part or all of their savings confiscated due to the mismanagement of governments.

this is why this fork is therefore important and creates a precedent, and that the process used in the shadows to pass this fork is dangerous, it is not only a question of the 10 million linked to the hack, but it is also a question of what could happen in the future, a precedent in a legal framework is extremely strong, a hard fork to modify the result of the past creates a precedent to which it is no longer possible for justice to say that it is not possible.

All the work of NolanV, whom I thank for having done it for free, highlights that there is a problem in knowing and measuring the reality of the risk mentioned above, as it stands for me it is impossible to know to what extent Gnosis LTD or affiliate could influence the chain on such a problem.

As already mentioned in another message, I have personal convictions which are the code and the law, but I am also realistic, this vision is not that of the majority of users.

I am not opposed to the fork, but I am against the development of the latter.
We are cruelly lacking in factual data on the reality of the decentralization situation, the problem is not being centralized or decentralized but being able as an individual to be able to measure the level, and make independent personal decisions with knowledge of the cause.

The other important point to understand is that there is a factual reality that a DAO is not a rapid body for making decisions and therefore in a devolution framework it is not always a good thing that the DAO decides on everything, sometimes you need a group or entity that can make decisions and be fired in the event of bad decisions, it is a more reactive and effective approach to accomplish the vision of the DAO.
So having a temporary centralization on a certain subject can be beneficial, but the DAO can still withdraw all rights to avoid various deviations from its objectives.
Unfortunately here this does not seem to have been the case, the DAO did not have a say, which shows that there is a real problem on the DAO’s side compared to the other major players.

Finally, given that there is no other choice on the table for the fork, we have been clearly told that it must take place on December 22, I regret that voluntarily or through oversights there was not this debate which was requested more than a month ago which could have allowed the fork to be carried out as planned with a clear framework.
Perhaps the implementation of security resulting from its framework decisions to activate a fork could not have been put in place at the same time but with a DAO which has a vision, a plan, trusted actors who have a mission, and the possibility of removing the trusted actors who do not meet the expectations of the DAO we could have moved forward with complete peace of mind.

We cannot change the past, the DAO and the Fork-reflections group cannot force the actors of the Fork-approve group to delay the fork.
It therefore seems that this creates a precedent which if it is not corrected by technical constraints in the future could pose serious problems, which on a personal level will make me review my position and exposure to Gnosis chain, but also the message that I will transmit and probably more widely triggered an important reflection at the level of RealT for our ecosystem.

In conclusion, it is important to understand that I am not saying that the fork in time as such poses a problem, that absolute immutability is required (immutability is my personal position but not the rational position), what is proposed is a solution which could be beneficial for the future of the chain if we take into account the desires of the mass of users and institutional actors, but my main concern here is how to ensure that this tool at the same time can like the nuclear can be devastating (like a nuclear bomb) if there is not all the necessary security, just as is the case with nuclear power plants and airplanes.
In the current situation I have the impression that we are going to do an experiment with a tool that could fall apart in our hands.

Thank you to those who managed to read everything, I have a dramatic and pecimistic tone, because we must not neglect the impact that this fork will have

4 Likes
60

I can’t help but think you’re going off on a tangent…

Just your quote from that paragraph:

  • On Europe, its politics, its democracy which, according to you, is no longer a democracy, its elected officials who, according to you, are not truly elected. Whereas they are elected in a democracy through a voting system, not a far-right populist one.

Then, moving on to pressure from individuals, crises of state bankruptcy, confiscated data, fears of data “not stolen as you say,” but “preserved”—to use the correct terminology; this leads straight to paranoid fear, which is a poorly handled aspect of your writing. And, above all, it has absolutely nothing to do with the topic.

Greece already had regulations involving the withdrawal and confiscation of data to get the country back on its feet. The EU is not a dictatorship but a democracy and a system of freedom. There are laws, of course, but if you have nothing to hide, the data is perfectly legitimate in your personal life. I think you’ve completely gone off-topic here, which is a real shame. So please avoid these kinds of long, pointless paragraphs of personal rambling about blockchain. Your personal opinion on this has absolutely nothing to do with it; it’s just another way to get off-topic and start talking about a thousand incomprehensible things related to blockchain. And being so pessimistic about nuclear bombs and confiscation makes it seem like you’re trying to protect the hacker at all costs—an impression you’re so keen on that you’re veering into apocalyptic and bleak political futures that have nothing to do with blockchain or the fork. It almost makes me doubt your personal involvement in this fund rescue, as if it’s really motivated by some personal interest.

And, from a decentralized perspective, if you want to talk about centralized blockchains, there are other well-known examples, like the one that managed to block the hacker’s wallet, causing them to lose their stolen funds. These funds were then returned to Balancer by the blockchain administrators. There are other blockchains that are far more centralized than the one involved in the fork and are doing just fine within their own systems.
Many people even feel more secure having large sums of money in these centralized blockchains because, in case of a hack, they know their funds can be recovered.
So, although Gnosis is a completely decentralized system, if we’re really talking about the fear of centralization in this area, it’s these other centralized blockchains that are relevant here. Discussing the future of rules is definitely not the point of this fork, and there’s certainly nothing wrong with criticizing the effort made to recover the funds from the chain.