Q3 2026 Security & Compliance$1.18B stolen across 107 incidents. Main drivers: supply chain, smart contracts, keys.
Get the report →

Hacken x Extractor Q3 2024 Web3 Security Report

29 min read

By Hacken

Quarterly report · Q3 2024
Q3 2024 Web3 Security Report
How much was stolen, and how much could have been saved?
General Trends
28The lowest number of hacks in years
$463.6MTotal value stolen
Recovery Challenges
94.9%Worst quarter in funds recovery
$440.1MTotal lost without recovery
Preventable Exploits
28.7%DeFi hacks could have been prevented
$25.6MPotential savings with Automated Incident Response
Attack Types
Access control is the most dangerous attack, with losses double all other attacks combined.
Smart contract vulnerabilities most commonly appear after the deployment of new versions.
Rug pulls have declined, but many scammers have shifted to memecoin platforms.
Attack Targets
CEXs are the largest pools targeted by hackers.
Bridges remain the frequent targets for attacks.
Highest share of value lost in Asia.
Preventive Measures
Take Contract Upgrades SeriouslyEnsure all smart contract updates are thoroughly audited and tested before deployment to prevent the introduction of new vulnerabilities.
Bug Bounties Are Better Than Forced BountiesProactively offer bug bounty programs to incentivize security researchers to report vulnerabilities responsibly, rather than exploiting them.
Implement Automated Incident ResponseDeploy real-time monitoring and automated response systems to detect and mitigate attacks swiftly, minimizing potential losses.
Enhance Private Key SecurityUse hardware wallets and secure key management solutions to protect private keys from unauthorized access and malware.

Global Observations

As we wrap up the third quarter of 2024, it’s time once again to review the security landscape of the Web3 industry. With increasingly complex technology and layering, it pays to remain vigilant and fully understand the attack vectors as well as the solutions to counter them.

At Hacken, we understand all too well the pain of those who lose digital assets. Liquidity plays a crucial role in most crypto protocols, and losing it is one of the worst things that can happen to a project.

That’s why we are pleased to inform you that despite a few high-profile attacks, which we will discuss later, this quarter has seen the lowest number of exploit cases not only this year but also in the past three years.

Number of Incidents per Quarter
95
107
113
107
132
127
98
67
41
28
Q2Q3Q4Q1Q2Q3Q4Q1Q2Q3
202220232024

A similar trend is observed in the amount of assets stolen, which has been steadily declining throughout the year.

Q3Q2Q1
Types of attacks%Money stolenNumber on incidentsMoney stolenNumber on incidentsMoney stolenNumber on incidents
Access Control68.71%$316,069,0008$397,291,00010$669,169,08924
Smart Contract Vulnerability9.12%$42,296,0009$0-$46,768,93017
Reentrancy7.27%$33,460,0003$0-$0-
Rug Pull0.53%$2,440,0002$3,690,0007$63,967,87915
Flash Loan Attack0.71%$3,256,0003$4,489,0005$33,381,77610
Phishing12.06%$55,473,6181$0-$12,917,5501
Other0%$00$84,358,00017$00
Oracle Issue2.30%$10,600,0002$23,100,0002$00
Total$463,594,61828$512,928,0041$826,205,22467

However, this is the worst quarter in recent times in terms of recovered or frozen funds. Among all the victims, only three projects were able to recover the lost assets. We had hoped that the trend of refunding a percentage of the siphoned funds, which was common in previous quarters, would continue—but alas!

Lost vs Recovered/Frozen FundsQ3Q2Q1
Total Lost$463,594,618$512,928,000$826,205,224
Total Recovered/Frozen$23,520,000$347,431,288$439,533,698
Lost without Recovery$440,074,618$165,496,712$386,671,526
Total Recovered/FrozenLost without Recovery
Q3
Q2
Q1

If we look at projects by region, most of the losses this quarter occurred in Asia.

RegionQ3Incidents Per RegionAmount Per Region
Asia3$264,000,000
Europe4$22,160,000
North America3$15,000,000
Australia1$43,300,000
Unspecified17$119,134,618

Apart from identifying and describing cases of crypto asset theft, we also explored what could be done to prevent larger losses using automated response tools. Our conclusion is that in the DeFi sector alone, at least 28.7% of assets could be frozen as an attack unfolds. The details will be discussed in the following chapters, but for now, let’s dive into analyzing what has happened over the last three months.

Attack Types

The total number of attacks saw a notable decline compared to previous quarters, reflecting a positive trend in mitigating the most common vulnerabilities. We attribute this to the Web3 industry reaching a critical mass of robust security solutions and best practices. However, certain attack types remained prevalent, requiring ongoing vigilance and proactive measures.

Smart Contract Vulnerability9
Access Control8
Reentrancy3
Flash Loan Attack3
Rug Pull2
Oracle Issue2
Phishing1
Other0
Total28

1. Access Control Attacks

As usual, the most damaging type of attack remains the scenario where a malicious actor gains control over the seed or functions, enabling them to arbitrarily withdraw funds from wallets or smart contracts.

With a total of eight cases and $316M stolen, the access control category accounted for more than double the percentage of assets stolen compared to all other attack types combined.

One notable exploit, not typically included in this report but involving an Access Control element, occurred with an unnamed individual using a DeFi Saver proxy. This address became the target of a phishing attack, leading to a change in proxy ownership and a subsequent loss of over $55 million. Although individual fund losses are generally outside the scope of such reports, this case underscores the critical importance of vigilance for anyone in the crypto industry.

Another significant case occurred in mid-September, when several alleged members of the North Korean Lazarus Group infiltrated DeltaPrime, a leveraged farming protocol. They managed to obtain private keys for liquidity pools on Arbitrum and used them to drain $6 million in assets. The well-known on-chain analyst ZachXBT linked the individuals who infiltrated the protocol to a group of pseudo-developers using fake IDs to support one another. This fake network is then leveraged to secure employment within a company, where vulnerabilities are exploited from the inside.

WazirX India Access Control Hack

The WazirX India Exchange hack stands out as the largest attack in Q3 2024, with $230 million lost. Despite employing a robust multi-party security system, the exchange suffered a breach due to unauthorized fund movements from their wallets.

WazirX utilized a Gnosis Safe multisig wallet requiring 4 out of 6 signatures for transactions. Five of the keys were managed by WazirX, while the sixth was held by Liminal, a digital asset custody provider. The attacker managed to manipulate the system, obtaining signatures from three WazirX signers and one from Liminal, allowing them to upgrade the wallet to a malicious contract and siphon off the funds.

In the aftermath, both WazirX and Liminal conducted independent audits, each clearing themselves of vulnerabilities. Grant Thornton's audit confirmed no compromise in Liminal's infrastructure, while Mandiant found no evidence of compromise on WazirX's devices. This led to speculation about an inside job, although no concrete evidence supports this theory.

2. Reentrancy & Smart Contract Vulnerabilities

One of the most persistent methods of extracting assets from a protocol is the reentrancy attack. This involves exploiting a loop in a smart contract's withdrawal function, allowing the attacker to withdraw funds multiple times. It is particularly devastating for protocols with liquidity pools.

Although this quarter saw only three reentrancy attacks, they resulted in over $33 million in losses across various assets.

Another significant category of attacks stems from coding errors, design flaws, or inadequate testing of smart contracts, leaving them vulnerable to exploitation. This attack has happened nine times this quarter, leading to nearly the same amount of losses—approximately $42.3 million. Notably, most of these attacks occurred following the deployment of new updates without undergoing proper security audits.

Vow Token
$1.23m Hack
The Vow hack involved a critical error during live testing on the mainnet. The team modified a rate-setting formula, which allowed an attacker to exploit the temporary change and mint excessive tokens. This occurred within the first minute after deployment, suggesting either a targeted vulnerability scan or potential insider involvement.
Minterest
$1.46m Hack
This hack exploited a reentrancy vulnerability within its smart contracts on the Mantle Network, resulting in a $1.4 million loss. The attacker initiated the exploit by taking out a flash loan, which allowed them to manipulate market exchange rates through the project's mUSDY contract. By leveraging the reentrancy vulnerability, the attacker repeatedly borrowed and converted tokens, accumulating excess mTokens that were then used to drain assets from the protocol. This attack not only exposed weaknesses in price manipulation and reentrancy but also emphasized the importance of comprehensive security audits to prevent such exploits.
Terra Luna
$6.5m Hack
This continually struggling blockchain was targeted by a sophisticated exploitation of its Inter-Blockchain Communication (IBC) system. The attacker deployed a malicious CosmWasm contract, exploiting the MsgTimeout function to mint and transfer tokens off the platform. This breach was facilitated by a previously identified and patched vulnerability, but because no proper quality control or audit was done to the new update, it overwritten the old patch and opened the identified vulnerability once more.

3. Rug Pulls vs Memecoins

The decrease in the number of attacks may also be attributed to the decline of traditional rug pull schemes. While this is a positive trend, an important caveat must be noted:

Rug pulls typically involve launching a product, accumulating liquidity in pools, and then removing assets through various means. This behavior is alarmingly similar to that of many memecoins that have emerged on the Base, Tron, and Solana blockchains, coinciding with the disappearance of typical rug pulls.

What is even more concerning is that on Pump.Fun, the largest Solana platform for creating memecoins, over 2 million coins have been launched in recent months, with only 89 of them achieving significant adoption and reaching a market cap of at least $1 million.

This suggests that many perpetrators of rug pull scams may have migrated to these coin factories, where they mimic legitimate activity.

Project Types Affected

With the prevalence of smart contract exploits during this period, it is unsurprising that the DeFi sector was the most common target. Protocols with liquidity pools as part of their architecture are frequently probed by hackers searching for vulnerabilities. Additionally, the complexity and various integrations with other projects often create security gaps, leading to significant capital loss.

Total Amount Stolen (USD)$463,594,618
CEX$295,300,000
Yield Aggregator$35,410,000
Bridge$28,600,000
Lending / Borrowing$19,616,000
L1$14,500,000
Oracle Service Provider$5,200,000
Restaking Protocol$2,000,000
Trading Bot$3,000,000
Token$2,926,000
DeFi Ecosystem$1,500,000
Gaming / Metaverse$69,000
Other$55,473,618

Even the largest protocols like Aave may leave its periphery contracts unprotected. This was exactly the case with ParaSwapRepayAdapter that was not included in any audits and exploited by a hacker for $56,000.

As noted in previous reports, the two categories that typically lose the most money are CEXes and bridges. This quarter, only three attacks targeted centralized exchanges, yet they accounted for nearly 64% of all funds lost.

Three bridges were hacked - PolyNetwork, LiFi Finance, and the Ronin Bridge, notorious for the largest bridge hack in 2022. This time, however, it netted only $28.7 million - or 6.2% - of total losses among all three.

Additionally, in the case of Ronin Bridge it was a white hat MEV bot who front-runned exploit transaction, extracting around $12 million in ETH and USDC. The funds were quickly returned, and Sky Mavis awarded the hacker a $500,000 bounty while announcing security improvements to the bridge. Notably, this exploit was introduced by a recent, unaudited update.

Affected Networks

Chains By Incident Category
Incidents Per NetworkAccess ControlReentrancySmart Contract VulnerabilityRug PullFlash Loan AttackPhishingOracle Issue
Ethereum175180111
Bitcoin11000000
Optimism21010000
Tron11000000
Polygon11000000
Mantle10100000
Base10001000
Terra Luna10010000
BNB Chain41001200
Arbitrum21010000
PolyNetwork10100000
Bittensor11000000
Scroll10000001
Solana10000001

We closely monitor the distribution of hacks across various blockchain networks, identifying patterns in attack frequency and type. Overall, vulnerabilities vary across networks, with Ethereum and BNB Chain being the most commonly impacted by a wider range of issues. This analysis serves two crucial purposes:

  • It highlights network-specific vulnerabilities, revealing which attack vectors are more prevalent on particular blockchains.
  • It enables developers and security teams to implement targeted protective measures for smart contracts deployed on these networks.

It’s important to remember that a single attack can target multiple blockchain. Yet, by tracking these trends, we can proactively enhance security protocols and raise awareness about blockchain-specific risks, ultimately contributing to a more robust and resilient decentralized ecosystem.

Incident Prevention in Real Time

Out of $89.8 million lost in DeFi exploits over the Q3, approximately $25.6 million (28.7%) could have been secured with automated incident response strategies.
On-chain monitoring & automated incident response are crucial in understanding and preparing for security threats each project faces.

After analyzing DeFi exploit patterns this quarter, we conclude that a significant share of the lost funds could have been prevented with an Automated Incident Response Strategy.

Automated Incident Response Strategy (in Web3) is a system that utilizes automated tools and predefined actions to identify and counteract security threats in the blockchain ecosystem. It requires a potential threats analysis, determination of the on-chain conditions that flag these threats, and a defined list of automated on-chain actions that should be executed when malicious/highly suspicious on-chain conditions are met.

Exploits are usually not single-transaction hacks but involve a series of transactions with preparation and exploitation phases. In the Aave periphery contract hack, all actions were executed in the same transaction, including the creation of the malicious contract. However, most exploits involve multiple transactions. For example:

  • attacks with preparatory transactions, where suspicious activity, such as funding from crypto mixers;
  • multi-step exploits, where malicious transactions target different pools one by one;
  • direct exploitation of Access Control through malicious Proxy Upgrade before withdrawing funds

We propose an Automated Incident Response Strategy that can deal with multi-transaction attacks by monitoring on-chain activities and triggering automated actions in response to suspicious or malicious behavior. The analysis will reveal how much could have been saved with this Strategy.

The Automated Incident Response Strategy (AIRS) can be implemented using various techniques and can differ based on the level of customization for each Web3 protocol, as well as the type of on-chain monitoring employed (e.g., monitoring a mempool versus confirmed transactions). In this case, we define a Simplified Automated Incident Response Strategy, where we focus exclusively on confirmed transactions as the on-chain conditions, and the adjustments are made based on general, well-known smart contract behavior patterns rather than highly customized configurations for each protocol.

Automated Incident Response Strategy In Action

Let's walk through hack cases and analyze them in a way of applying Automated Incident Response Strategy.

Nexera

Nexera$1.5m hack
Incident response: 1.5M out of 1.5M Secured
100% Secured

As the NXRA token contract is upgradable, after pausing the contract, the team managed to burn the frozen tokens on behalf of the attacker's account. In that way, they did not let remaining NXRA tokens to be exchanged for liquid assets such as ETH.

Percentage of the total loss which was possible to secure: 100% ($1.5m)

The exploiter accessed ProxyAdmin contract via the exposed Private Key and upgraded several proxy contracts
Then, the attacker managed to withdraw a total of 47.2M NXRA tokens.
Attacker succeeded at swapping 15M NXRA tokens before the token got paused by the team. The rest of the NXRA tokens remained frozen on the attacker’s wallet

As the NXRA token contract is upgradable, after pausing the contract, the team managed to burn the frozen tokens on behalf of the attacker's account. In that way, they did not let the remaining NXRA tokens be exchanged for liquid assets such as ETH.

The liquidity in ETH that the attacker managed to withdraw from DEX swapping NXRA tokens is ultimately lost. However, it was possible to prevent this if the Automated action of Pausing the token was configured WHEN the Proxy is maliciously upgraded. Thus, there is no chance for the attacker to swap NXRA for ETH as the token is paused and cannot be transferred.

Penpie

Penpie$27m hack
Incident response: 11.1M out of 27M Secured
41.1% Secured58.9% Loss

The attacker exploited a reentrancy vulnerability by creating valueless versions of Pendle’s yield-bearing tokens and linking them to valuable assets.

There were 3 major attacks executed which took place on the ETH network. The biggest amount of assets stolen was in the first attack transactions, where the attacker managed to withdraw 15.7m out of 27m total drain.

Percentage of the total loss which was possible to secure: 41% ($11.1m); second and third attack would not happen in that way

attack №attack txexecuted byattack timeassets stolenworth $usd
attack 10x56e09Exploiter 206:23:35 PM UTC2,722.90 wstETH, 2,522,685.57 sUSDe, 1,373.13 agETH, 893.52 rswETH$15.7m
attack 20x42b2eExploiter 106:37:59 PM UTC1,367.72 agETH, 901.86 rswETH$5.6m
attack 30x663b5Exploiter 106:42:35 PM UTC1,359.65 agETH, 899.74 rswETH$5.5m

Contracts were paused 20 minutes after the first attack was executed.

Another key takeaway here is attackers deployed 5 attack contracts, only 3 of them were used in attacks, the 4th attack contract was deployed just 1 minute after Pendle paused their contracts. This fact underscores the importance of timely executed actions in order to respond to the ongoing incident.

The loss of $11.1m was possible to prevent, if Automated action of Pausing contracts was configured WHEN One of the Pools is drained for >50%, which was the case for first attack.

Ronin bridge

Ronin bridge$12m hack
Incident response: 2M out of 12M Secured
16.7% Secured83.3% Loss

Hack occurred due to the issue with the Bridge upgrade. The vulnerability was that the contract was not properly initialized, leaving a critical parameter empty. This oversight allowed attackers (which happened to be white-hat) to withdraw 2M USDC and 4,000 ETH without proper signatures.

Percentage of the total loss which was possible to secure: 17% ($2m)

The timeline of the exploit
9:37 AM UTC4K ETH withdrawn
10:11 AM UTC2M USDC withdrawn
10:15 AM UTCContract pausedThe Bridge was paused 4 minutes after the second withdrawal and 38 minutes after the initial one.

The withdrawal of 2m USDC was possible to prevent*, if Automated action of Pausing contracts was configured WHEN Bridge executes suspicious withdrawal of huge amount of liquidity**

* Here the withdrawal of 2m USDC counts to a total loss, even though it was executed by the white-hats, as it is still assets which were withdrawn as a result of vulnerability.

** The definition of “suspicious withdrawal” refers to the fact that this particular withdrawal of 4K ETH was not a regular one in terms of function call parameters. The fact that it also was of 4K ETH (which is the limit for single-transaction withdrawal) might be determined/flagged as a highly suspicious (or even malicious right due to these facts) transaction. This note is made due to the non-binary nature of the evaluation of malicious transactions.

Onyx

Onyx$3.8m hack
Incident response: 100% Secured
100% Secured
  • The attacker exploited a known precision vulnerability in the Onyx Protocol's forked CompoundV2 code base, which allowed manipulation of the token exchange rate.
  • Once the exchange rate was manipulated, the attacker withdrew approximately 3.8 million VUSD.

Percentage of the total loss which was possible to secure: 100%($3.8m)

Breaking down the timeline of the Onyx exploit, the attacker withdrew 3.8 million VUSD and began exchanging VUSD for ETH just six minutes later. Sixty minutes after the attack took place, the last VUSD to ETH swap was executed.

However, the VUSD stablecoin token contract includes a blacklist functionality, but the exploiter was not blacklisted within those 60 minutes. The 3.8 million VUSD represents 7% of the overall VUSD circulating supply, and the withdrawal of such a percentage is a significant anomaly that should trigger immediate alerts. Upon detecting such an unusual transaction, the system could automatically invoke the blacklist function to prevent the attacker from moving or swapping the stolen VUSD, preventing a significant depeg.

VUSD price, 24 to 30 September 2024, CoinMarketCap chart

These case studies demonstrate the potential effectiveness of an Automated Incident Response Strategy in mitigating losses during security incidents. While AIRS cannot prevent all types of attacks, it can significantly reduce the impact by enabling automated responses to suspicious activities.

Disclaimer:

It is important to note that the proposed Threat Prevention and Incident Response Strategies are simplified versions of the Strategy adjustment for each of the projects. In reality, protecting Smart Contracts in real time requires thoughtful sophisticated adjustments to each of the Smart Contracts and to the system of the Smart Contract as-is, modeling threats and using different monitoring approaches.

Preventive Measures

1. Audits and Bug Bounties

Despite the losses this quarter, we want to emphasize that the overall security health of the Web3 industry has improved significantly. Avoidable mistakes are less frequent, and companies are generally adhering to best practices for securing their infrastructure. Crypto is maturing, and we hope this positive trend continues.

That’s why it’s particularly frustrating to see companies making mistakes like deploying updates without proper security audits, only to be hacked within weeks and then seek an audit afterward.

Audits

Audits do make a difference. While they are not foolproof and can sometimes miss issues—as happened with Convergence Finance, one of two instances this quarter where an audited smart contract was hacked—they greatly reduce the likelihood of exploitation in Web3 projects.

Bug Bounties

Same can be said about bug bounties. Whenever a hack happens, companies often are forced to offer large bounties to return the funds. Each hack is a reputation hit, while having an active bug bounty with clearly defined scope and rewards usually increase a company's reputation in the eyes of their community

Audits
Security MeasuresQ3out of
Audited Projects1127
Not Audited1627
Smart Contract in Audit Scope211
Bug Bounties
Security MeasureHacked ProjectsTotal Hacks
Active Bug Bounty During Attack728
Vulnerability in Bug Bounty Scope128

2. Automated Incident Response Strategies

As we analyze the exploits from this quarter, a clear pattern emerges: many attacks could have been mitigated or prevented through automated incident response strategies. Our analysis indicates that implementing such strategies could have averted at least 28.7% of the DeFi total losses in Q3

Most exploits follow a sequence of transactions, including preparation and exploitation phases. While some, like the Aave periphery contract hack, were executed in a single transaction, the majority involved multiple steps that could have been detected and addressed automatically.

Automated Incident Response Could Have Averted These Hacks
100%
of the loss in Nexera's
$1.5m hack
41%
in Penpie’s
$27m hack
17%
in Ronin bridge’s
$12m hack
100%
of the 3.8m VUSD loss in Onyx could have been contained by blacklisting the attacker
50%
of the loss in DeltaPrimte’s
$6m hack
85%
of the loss in Shezmu
$4.9m hack

Hacken Extractor offers an on-chain monitoring service that supports all of the Automated Risk Mitigation and Incident Response Strategies described above, as well as more Sophisticated Threat Detection and Prevention Strategies adjusted specifically to each project in the Web3 space.

Protect Against Social Engineering, Malware, and Phishing Attacks

Enhance Private Key SecurityUse hardware wallets and secure key management solutions to protect private keys from unauthorized access and malware.
Educate Team Members on Security Best PracticesRegularly train staff to recognize and avoid phishing attempts and social engineering tactics that could compromise credentials.
Implement Multi-Factor Authentication (MFA)Require MFA for all sensitive accounts and systems to add an extra layer of security beyond just passwords or private keys.
Conduct Regular Security Audits and Penetration TestingPeriodically assess systems for vulnerabilities through professional security audits and penetration tests.
Limit Access PrivilegesApply the principle of least privilege by granting employees the minimum levels of access—or permissions—needed to perform their job functions.
Establish Incident Response PlansDevelop and maintain a comprehensive incident response plan to ensure quick and effective action when a security breach is detected.
Use Anti-Phishing TechnologiesImplement email filtering and web security solutions to detect and block phishing attempts before they reach employees.

Future Predictions

As we conclude our analysis of Web3 security failures in Q3 2024, it’s clear that the industry has made significant progress in mitigating vulnerabilities and reducing the frequency of successful attacks. Despite a few high-profile incidents, this quarter saw the lowest number of exploit cases in the past three years, along with a decline in the total value of assets stolen.

However, vigilance and proactive security measures remain crucial. Access control attacks continue to be, and likely will remain, the most damaging type of exploit, accounting for a disproportionate share of stolen funds.

On a positive note, we expect the amount of funds returned or frozen to remain high, even as the overall number of attacks decreases. There are now more tools available to facilitate this, along with better capabilities to notify exchanges where stolen funds could be laundered, and a regular practice of rewarding white hat hackers after the recovery of funds.

Looking ahead, we anticipate a decrease in both the frequency and scale of bridge hacks, thanks to the adoption of newer, more secure architectures. Nonetheless, it is essential for projects to prioritize comprehensive security audits, maintain active bug bounty programs, and implement automated incident response strategies to further mitigate risks.

As the Web3 industry continues to evolve, collaboration, knowledge sharing, and a proactive approach to security will be key to building a more resilient and trustworthy ecosystem. By learning from past incidents and adopting best practices, we can pave the way for a future where the benefits of decentralized technologies are realized without compromising the safety of users’ assets.

Subscribe to our newsletter

Be the first to receive our latest company updates, Web3 security insights, and exclusive content curated for the blockchain enthusiasts.

Speaker Img

Tell us about your project

Follow Us