Cronos rolled back its chain to undo a $75 million hack, which is chilling

Cronos verifier rolled back more than 10,000 blocks, reversed the Tectonic attack, and froze assets to save US$69 million, but sparked a heated debate: Can a blockchain that can be rewound at will be called a blockchain?

Cronos verifier suspended block production on August 30, rolled back more than 10,000 blocks, restoring the chain state to before the attack, thereby erasing about two hours of transaction history for all users on the network.

The Tectonic attackers used approximately $600,000 to push the price of TONIC tokens 100 times in 20 minutes, then deposited 364.6 trillion of the high-priced tokens as collateral and lent approximately $75 million from the loan agreement.

Only about US$6 million was transferred to Ethereum before the chain was suspended; the remaining US$69 million was frozen at the Cronos address until a rollback operation completely erased the attack transaction from the main chain.

Tectonic's total locked position value plummeted from US$121.7 million to approximately US$3 million in the 48 hours after the attack, a drop of as much as 97.5%.

RedStone's co-founder said its oracle report was accurate and blamed Tectonic's collateral control mechanism, saying that only one parameter (which enforces a liquidity-related borrowing cap) could have prevented the attack.

On August 30, Cronos did something that most blockchains claim cannot and will never do.

Its validators collaborated to implement an emergency pause, agreed to discard more than 10,000 blocks of established history and restart the chain from a snapshot. The snapshot was timed before a loan agreement called Tectonic lost $75 million due to a collateral-manipulation attack. The stolen funds, except for the approximately $6 million that had been transferred to Ethereum, no longer existed on the chain after the restart.

This response worked. It controlled the loss. It likely saved depositors from losing all of their assets in Tectonic.

However, it also raises a question that the industry has been avoiding since Ethereum's DAO fork in 2016: If a small group of validators can rewrite the history of the chain to reverse theft, what is the difference between this chain and a cumbersome database? The answer to this question is even more important today than it was in 2016, when the industry has been preaching to agencies, regulators and retail users that blockchain provides features that traditional financial infrastructure does not possess: transactions that cannot be reversed by any single authority. Cronos just proves that this statement is not universally true.

How Tectonic lost $75 million in 20 minutes

The pattern followed by this attack has been so well documented that DeFi security researchers named it: Mango-style pull-and-loan attack.

Tectonic is the largest loan agreement on Cronos, with a total locked value of approximately US$121.7 million and an active loan amount of US$82.7 million. The agreement allows users to deposit their governance token TONIC as collateral. TONIC's collateral factor is 20%, which means users can borrow up to one-fifth of the reported value of their collateral. This parameter assumes that the price reported by TONIC is close to its actual liquidation value. But this is not the case.

The attacker spent approximately $600,000 buying TONIC on the thinly liquid Cronos market, pushing the token price up approximately 100 times in approximately 20 minutes. Subsequently, using this hyped valuation, the attacker deposited 364.6 trillion TONIC with Tectonic, creating a mortgage position reported to be worth approximately $375 million. With this false collateral, the attackers lent approximately $75 million in liquid assets from other depositors.

The numbers clearly illustrate the problem: $600,000 in investment resulted in a withdrawal of $75 million. The return on capital is approximately 12400%. The collateral supporting the loan would not be worth even a fraction of its reported value if it were tried to liquidate, because a sell-off would instantly return the price to normal. Tectonic's lending market was hollowed out by its own pricing assumptions.

Prior to the attack, Tectonic held nearly half of the capital deposited in all Cronos DeFi apps. Within 48 hours, its total locked position value plummeted from $121.7 million to approximately $3 million. The project, which was supposed to be the cornerstone of the Cronos DeFi ecosystem, has become its most expensive liability.

Pause: The validator pulls the emergency brake valve

The Cronos validator discovered the attack within minutes and made a decision that a truly decentralized network could not make quickly: They stopped block production.

This pause froze everything. It's not just Tectonic. Every transfer, every smart contract interaction, and every cross-chain bridge transaction on the entire Cronos network has stalled. Users who have no connections to Tectonic cannot transfer their funds. The cross-chain bridge connecting Cronos to Ethereum and other chains stopped processing. RPC providers that serve applications on Cronos are also in the dark.

Timing is crucial. When the verifier stopped block production, the attacker had successfully transferred approximately $6 million to Ethereum-a place where the Cronos verifier had no jurisdiction. The remaining $69 million remained at the identified Cronos address, which was frozen but was technically still under the control of the attacker on the suspended chain.

Kris Marszalek, CEO of Crypto.com, posted that the exchange and App are operating normally and "all funds are safe." The statement specifically refers to assets held through Crypto.com's centralized service, not funds deposited with Tectonic. This distinction is crucial. Crypto.com is closely associated with Cronos, but Tectonic runs as a stand-alone decentralized application. The failure of one system does not necessarily endanger another, and Marszalek's assurance only covers the centralized end.

Rollback: Erasing 10,000 blocks in the history of tens of thousands of people

Instead of restarting from a suspended state and hoping to freeze the attacker's address through governance or technical intervention, the Cronos verifier chose the ultimate solution. They restored the chain to a snapshot state it had before the attack, rolled back more than 10,000 blocks, and restored block production from a block height of 90,896,189.

Attack transactions no longer exist on the main chain. At the same time, all other transactions that occurred in those erased blocks disappeared. Legal transactions, token transfers, contract deployments, and anything that happened to occur during that approximately two-hour window disappeared.

Cronos described the suspension as a "validator consensus urgent action" to protect users. The post-mortem analysis report that the chain has promised to release but has not yet been released should explain the specific process through which verifiers reached agreement on recovery points. What we currently know is that decisions were made quickly, executed by a small group of validators, and reversed the established history of a public blockchain.

Tatum, an infrastructure provider that serves Cronos developers, had to replay all chain data starting in block 90, 896, 188 to re-synchronize its systems. Other RPC providers, browsers, and cross-chain bridges require similar reset operations. This rollback not only affected the attacker, but also forced all services connected to Cronos to adapt to a new real-world version.

Why the oracle is not the problem

After a price-fixing attack, people's instinctive reaction is to blame the oracle. RedStone co-founder Marcin Kazmierczak refuted this claim in a statement to crypto.news.

He said: "There is nothing wrong with the oracle. It accurately reported the price of TONIC in the liquidity pool that it read at the time." This distinction is crucial. Even if prices are artificially pushed up, a oracle that reports current market prices is fulfilling its duty. The failure lay in the agreement itself, which accepted the reported price as a safe basis for borrowing without checking whether the token could actually be sold at that valuation.

Kazmierczak points to a missing security measure: a borrowing cap tied to enforceable liquidity. Such a cap would limit the amount of borrowing based on how much collateral could actually be sold without triggering a price collapse. Even if TONIC's reported value soared 100 times, a properly set borrowing cap would limit borrowing to what the market can absorb.

He said: "Reporting a price and verifying whether it is safe to lend at that price are two different things, and Tectonic's design confuses them." He rejected the idea that using a longer weighted average price window would deter attacks. He believes that a 100-fold increase in prices in 20 minutes is not a volatile event that can be resolved through smoothing. This is a signal that the asset should not be used as collateral at any meaningful scale.

Such attacks are not new. That's the problem.

The tactics used by the Tectonic attackers were almost identical to the one Avraham Eisenberg used against Mango Markets in October 2022: hyping illiquid MNGO governance tokens, then borrowing liquid assets, taking away more than $100 million. A Manhattan jury found Eisenberg guilty of commodity fraud, commodity manipulation and wire transfer fraud. A federal judge later quashed the convictions citing venue problems and insufficient evidence for wire fraud charges.

The significance of the Eisenberg case is more than just technical similarities. His legal defense is that the rules of the agreement allow his actions, that smart contracts operate as designed, and that exploiting design flaws does not amount to fraud. The jury disagreed, but the decision to quash the conviction left the legal status of this attack vector open. Anyone copying this strategy today is in a true legal gray area, which may partly explain why such attacks persist.

Just three days before the Tectonic attack, an attacker used the exact same tactic to steal $8.7 million from the Moonwell protocol on the Base chain against extremely illiquid MAMO tokens. Moonwell responded by lowering the borrowing limit in Base's core markets to 1 wei, effectively shutting down new lending operations. This fix existed before the attack occurred. But the agreement was not implemented until after the damage was caused.

Moola Market on the Celo chain also suffered losses due to the same model in October 2022 (the same month as Mango Markets). Four years later, the attack still works because the financial incentives provided by listing governance tokens as collateral outweigh the perceived risks. Agreement teams benefit from higher TVL numbers. Governance token holders benefit from greater practicality. The cost of weak collateral parameters has been hidden until someone tests whether the market can absorb sudden selling of collateral tokens. The result was no. Never. For these low-market governance tokens to be safely used as collateral under their set mortgage factors, the liquidity required does not exist at all.

After another security incident on August 25, the Cosmos EVM chain was asked to be suspended. KiiChain reported that 148.3 million KII tokens were lost through 18 attacks. MANTRA suspended its network a few days ago to investigate another incident. Three chains were suspended within a week. Given this frequency alone, anyone who views ultimate certainty as an attribute of its blockchain should be concerned.

A comparison of the DAO fork and its inapplicability

The 2016 Ethereum DAO fork is an obvious precedent. An attacker used a reentry vulnerability to steal approximately $60 million from The DAO (at current prices), and the Ethereum community voted to hard fork, creating a new chain that reversed the theft and an original chain that retained the established history (Ethereum Classic).

This comparison is instructive, but differences are more important than similarities.

The DAO fork went through weeks of public debate. Discussion posts on CoinDesk, Reddit and Bitcointalk have thousands of comments. Miners use their computing power to vote. The community was divided, and the Ethereum classics were born as a permanent monument to the principle of "code is law". The process was so painful that Ethereum later regarded immutability as almost sacred. The Ronin Bridge lost US$625 million in 2022. The Wormhole Bridge lost $320 million that year. No one has seriously proposed rolling back Ethereum for these two things.

Cronos did something similar within hours, with a handful of validators. There is no community vote. There were no weeks of debate. There is no chain division. Forks that do not preserve the original history for dissidents to use. The validators agree, roll back, and move on. The problem is speed, because a rollback that requires broad community consensus and weeks of deliberation is a last resort, whereas a rollback that a small group of validators can perform in hours is a management tool. Management tools are used.

Centralization of verifiers explains this speed. Because the Cronos chain is maintained by relatively few validators, many of which are controlled by or closely associated with Crypto.com, the number of independent parties that need to agree to coordinate pauses and rollbacks is far less than Ethereum, Bitcoin, or any chain with a large and diverse collection of validators or miners. This is not a flaw in the response to the Tectonic attack. This is the structural condition that makes this response possible.

As one critic put it: If $75 million is worth rolling back, what about $50 million? What about $10 million? And, apart from hacking, what types of events are sufficient for a verifier to press the "reload" button? In the absence of a published governance framework on when to roll back, the answer is what the validator set decided at the time. This is not decentralized governance. This is discretion, and discretion without rules is power.

Who lost money in erased blocks

This rollback took control of the attack. But it also eliminates legal activities.

During the approximately two-hour window between the attack and the chain was suspended, every user executing a transaction on Cronos was reversed. Transactions on the decentralized exchange were cancelled. Token transfers between wallets are voided. Smart contract interactions that have nothing to do with Tectonic are erased from the main chain as collateral damage to state restoration.

Cronos has not announced how many non-attack transactions were lost in this incident. More than 10,000 erased blocks represent approximately two hours of Cronos network activity at its normal throughput. For a chain that has recorded more than 100 million transactions and supported more than 500 developers since its launch, even two hours represent a lot of legitimate operations.

This asymmetry is shocking. Tectonic depositors who lost funds as a result of the attack had their balances restored to pre-attack levels. However, anyone who completed a legal transaction, deposit or withdraw cash during the erasure window was void without receiving any compensation or even recognition.

This creates a strange incentive. If you are robbed on Cronos, the verifier may rewrite history to make up for your loss. If your legitimate transaction happens to fall into the reach of someone else's hacking, you will lose it. This rollback optimizes the remedy for one type of damage, but creates another type of damage.

No set of validators has explained how they weigh these conflicting interests. Cronos's post-mortem report should cover this. Whether it will be, is another question.

What rollback means for builders on Cronos

Developers building applications on Cronos now face a design constraint that did not exist before August 30: Any state of their application creation may be retroactively erased by the verifier consensus.

For a simple token exchange, the consequences are annoying but controllable. Users can resubmit. For applications that interact with external systems, the impact is more serious. A payment processor that confirmed the Cronos transaction and shipped it would have no recourse if the transaction was later rolled back. A oracle that pushes data to Cronos and triggers other on-chain operations based on confirmation messages cannot cancel these triggered operations.

The problem is more complex for protocols that span multiple chains. If a user makes a deposit to Cronos and that deposit triggers coin on another chain, a rollback of Cronos removes the deposit but not the coin. Cross-chain state becomes inconsistent, and the responsibility for coordinating resolution falls on the protocol team rather than on the validator ordering the rollback.

Tatum's response illustrates infrastructure costs. The company had to replay all chain data from the recovered block to get its API resynchronized. Each indexer, subgraph, and data service that tracks Cronos faces the same resynchronization burden. For infrastructure providers operating on dozens of chains, supporting a chain that may roll back at any time increases operational costs that chains with trusted finality do not impose.

Trump Media's CRO Treasury joint venture with Crypto.com (terminated August 7) had proposed using Cronos to tokenize assets. If the transaction continues until the time of the Tectonic attack, this rollback will erase the tokenized equity positions. This scenario alone is enough to make any real-world asset tokenization project think twice before choosing a chain where a verifier can rewrite history.

$6 million: The number that proves limitations

The $6 million that the attacker transferred to Ethereum before the chain was suspended survived the rollback. It is on a chain that cannot be reached by Cronos verifiers.

This is the physical limitation faced with every rollback. The authority of a blockchain ends at its own boundaries. Once value is transferred to another chain, the consensus rules for the receiving chain take effect. Ethereum's validators have not agreed to Cronos's rollback and are not obligated to respect it. The attacker's balance on Ethereum is final, while their balance on Cronos is not.

This gap is critical for anyone building cross-chain applications on Cronos or similar networks. If a chain can be rolled back, any value that has not left the chain before the chain is suspended risks being erased. Cross-chain bridges become escape pods, and bridging speed becomes a safety attribute unexpected by protocol designers.

The attacker knows this. The first reaction to stolen funds is to transfer them to Ethereum. The approximately two-hour window from the time the attack occurs until the chain is suspended is a race between the attacker's bridge speed and the validator's coordination speed. The validator won most of the victories. But $6 million is not a small amount.

Things to Watch

Release of the Cronos post hoc analysis report. The Verifier Set promises to provide a comprehensive explanation of the attack, suspension decision, rollback process and restart. Before the document was released, the community was unable to assess whether sufficient safeguards existed or whether the rollback followed any clear governance process.

Tectonic's total lockup value and how depositors handle it. The total value of locked positions has plummeted from $121.7 million to $3 million. Will depositors receive compensation, recovery plans, or nothing, which will show how Cronos handles the failure of agreements within its ecosystem.

CRO's price behavior after rollback. A set of validators that can rewrite history should have a higher governance discount on their transactions compared to chains that cannot do this. Whether the CRO reflects this discount will show how the market prices immutable risk.

Whether other chains adopt a rollback strategy. The MANTRA, Ontology and Cosmos EVM chains have all recently been suspended. If any of them used Cronos as a precedent for state rollback, this practice could become normalized in smaller chains.

Adoption of borrowing caps in the DeFi Loan Agreement. RedStone's Kazmierczak points out the fix. Whether the agreement implements it or continues to list low-liquidity governance tokens without a borrowing cap will determine how often such precise attacks occur again.

FAQs

What happened to Cronos on August 30?

Cronos verifiers suspended block production after attackers stole approximately $75 million using Tectonic, the chain's largest lending agreement. Subsequently, the validator rolled back more than 10,000 blocks, restoring the chain to its pre-attack state, and erasing the attack transactions from the main chain history.

How did Tectonic attackers steal $75 million?

The attacker spent approximately $600,000 to push up the price of Tectonic's governance token TONIC approximately 100 times in 20 minutes. The attackers then deposited 364.6 trillion of hyped TONIC as collateral and borrowed $75 million in liquid assets from other depositors. The attack took advantage of a 20% mortgage factor set by Tectonic on a token with little actual liquidity.

Did Cronos 'rollback recover all stolen funds?

No. About $6 million had been transferred to Ethereum before verifiers suspended block production. These funds exist on Ethereum and Cronos Verifiers have no jurisdiction over them. When the validator restored the chain to its pre-attack state, the remaining $69 million was effectively erased.

Is Cronos the first blockchain to be rolled back after a hack?

No. The most famous precedent was Ethereum's DAO fork in 2016, reversing approximately $60 million in stolen funds. The key difference is that the Ethereum fork took weeks of debate and community voting, while Cronos was rolled back within hours by a small group of validators without a public vote.

What is a Mango-style pull-and-loan attack?

Named after the 2022 Mango Markets attack, this attack method is to hype a illiquid governance token, deposit it as collateral in a lending agreement, and then borrow liquid assets using the hyped valuation. Assets lent are real and liquid; collateral is not. Tectonic and Moonwell were attacked in this pattern within three days in August 2026.

Could Tectonic attacks have been prevented?

RedStone co-founder Marcin Kazmierczak thinks it can. A borrowing cap tied to enforceable liquidity limits the amount borrowed regardless of TONIC's reported price. The oracle reported the correct market price. The agreement failed in that it accepted that price as a safe basis for borrowing without checking whether the token could be sold at that valuation.

What does Cronos rollback mean for other blockchains?

During the week at the end of August 2026, three separate blockchains were suspended: Cronos, Cosmos EVM chains, and MANTRA. If Cronos rollback is seen as a successful response, a small chain with a centralized set of validators may adopt the same approach, possibly normalizing state reversals as a security tool.

Should I continue to hold funds on Cronos?

This is educational analysis, not investment advice. This rollback shows that Cronos verifiers can and will change the history of the chain to control damage. Whether this makes the network more secure or less trustworthy depends on whether you value the ability to reverse theft or the ultimate certainty of the transaction. Assets transferred to other chains before the chain was suspended are not affected by Cronos rollback.

Disclaimer: This article is for reference only and does not constitute investment or financial advice. All quoted data are accurate as of September 2, 2026. The information presented here reflects publicly available data and signed statements. Readers should study for themselves before making any financial decisions.