Cosmos EVM Flaw Exploited After Widespread Blockchain Exposure Was Confirmed It has been revealed that Cosmos Labs has disclosed exploiting a critical vulnerability in its shared Cosmos EVM module, which resulted in the exploitation of funds from six blockchain networks between August 20 and August 25, 2026. GHSA-7g4w-cg88-2cq2 affects Cosmos EVM versions below 0.6.2 and versions between 0.7.0 and 0.7.1  Cosmos Labs has classified the issue as Critical, however the advisory is not accompanied by a CVE, weakness classification, or CVSS score. In the versions 0.6.2 and 0.7.2 patches, state-breaking changes were introduced, so affected chains were required to perform coordinated network upgrades. Cosmos Labs has advised operators that if they cannot complete an immediate upgrade, they should cease operations, rather than proceed with governance-based upgrades while remaining exposed.  A vulnerability was reported by the company on April 25 through a bug bounty program. Following a failure by Cosmos Labs to reproduce the issue on networks using 18-decimal configurations, the team initially concluded that the issue did not pose a threat to funds on live networks. It turned out that assessment was incorrect as the team established on August 13 that the vulnerability affected Cosmos EVM chains regardless of their decimal configurations.  After the flaw was discovered, Cosmos Labs developed a public silent-patch process, which was designed to address vulnerability that does not present a risk of fund loss on production networks. The patch was released on August 19, one day prior to the first reported exploit. It has since been revealed that several affected blockchains have halt their chains following successful attacks on their funds. KiiChain and TAC have halted their chains as a result, while MANTRA has temporarily closed its mainnet and later restarted it after the fix has been applied.  Three Networks Confirm Cosmos EVM Exploitation A security incident was reported by MANTRA among the first networks. The team initially described the incident as a result of an upstream dependency, which led to a halt of the chain. The team said user balances were not impacted, while two wallets controlled by MANTRA were compromised. The vulnerable component was identified by MANTRA as the Cosmos EVM module and reported that version 8.4.0 had been released to address the issue, resulting in the normal block production process returning.  In August, KiiChain confirmed a direct exploitation of the vulnerability. The attacker reportedly utilized the same technique 18 times, draining 148,326,583.15 KII before the network was halted at block 9,355,723 by validators. According to KiiChain, the vulnerability can only be found in the Cosmos EVM module rather than in its own chain-specific implementation of the Cosmos EVM module. This network has been suspended and is expected to resume via coordinated binary upgrades at a predetermined block height, with validators applying the new versions simultaneously.  TAC also suspended its network at block 24,671,475 on August 22 after an attacker drained funds from a single account. The team attributed the vulnerability to the Cosmos EVM module instead of TAC-specific code. Cosmos Labs has informed affected teams that they should contact their security contact, and a more comprehensive incident report will be released once the situation has been resolved. The company has not disclosed what caused the vulnerability.  Flaw Allowed Balance Manipulation Across Affected Chains This vulnerability is associated with the way Cosmos EVM reconciles Ethereum-style account balances with Cosmos SDK's x/bank module. The EVM tracks the balance of an account, and vesting accounts may also contain funds that remain locked. Staking functions are still capable of delegating locked funds.  An underflow of balance could occur if a vesting account delegated more than its available spendable balance. In lieu of rejecting the transaction, the calculation wrapped the balance into an extremely large UINT256 value. Cosmos EVM was then able to interpret the altered balance as newly created funds during reconciliation. The same mechanism could also be used against another account. Attackers could transfer artificially inflated amounts to victims' accounts, resulting in the subsequent reconciliation process burning their legitimate balances.  By creating vesting accounts without permission, the attack was successful. Using a contract deployed at a predetermined address, the attack was completed in a single transaction. This was a result of an underflow check introduced in May that was then backported to the release branches affected by the issue. The handling of locked balances has also been corrected, and a separate safeguard has restricted modifications to module accounts involving balance modifications.  Patch Distribution Drew Scrutiny Along with the technical issue, Cosmos Labs' handling of the vulnerability has become a major concern. Cosmos Labs confirmed on August 13 that the issue affected Cosmos EVM networks regardless of their decimal configuration, however, a silent patch process was used to apply the patch.  Shortly after the patched releases became available on August 19, a public pull request was made in Push Chain's Cosmos EVM fork, describing the vulnerability and its exploitation method. A less than 12-hour period later Cosmos Labs reported the first reported attack on MANTRA. It is believed that six networks have been exposed to the exploit. On the basis of information provided by the affected chains, approximately $2.87 million worth of assets were traded through decentralized exchanges, while another $2.85 million was traded through centralized exchanges. The figures have not been independently audited.  Cosmos Labs reports that 11 Cosmos EVM deployments that had not registered security contacts were discovered during the incident. The incident has also highlighted difficulties in reaching downstream operators. Due to the fact that there are over 115 public blockchains in the Cosmos ecosystem and no database of the networks that run its software, distributing urgent security updates remains a complex effort.

  The Drift Protocol, widely considered to be the largest perpetual futures exchange operating on the Solana blockchain, became the focal point of a highly coordinated attack on April 1, 2026, which is rapidly turning into one of the most significant breaches in decentralized finance this year.  In addition to revealing a vulnerability within one platform, this incident highlighted the sophistication of threat actors operating throughout the crypto ecosystem, which has increased over the years. Elliptic estimates that approximately $286 million was siphoned during the attack, with a pattern of transactions, asset movements, and laundering processes that resembled operations previously attributed to North Korean state-linked groups.  The breach would represent the eighth incident of this type recorded during the current year alone, contributing to a cumulative loss of over $300 million, should attribution be formally established. In general, it is indicative of the persistence of a strategic campaign in which upwards of $6.5 billion in cryptoassets have been exfiltrated in recent years activity that has been repeatedly linked to the financing of the country's weapons development programs by U.S. authorities. According to Elliptic's analysis released on Thursday, the $285 million exploitation event has multiple layers of alignment with operational patterns traditionally associated with North Korea's state-sponsored cyber units, making it the largest recorded incident this year.  Not only is the sequence of transactions on the blockchain highlighted in the assessment, but also obfuscation techniques are systematically employed, including staging asset dispersal and laundering pathways that mimic prior state-linked campaigns. As well as telemetry and interaction signatures, network-level interactions strongly suggest that a coordinated, well-resourceful intrusion is more likely than an opportunistic one. In response to the incident, Drift Protocol's native token has declined by more than 40 percent, trading near $0.06. This reflects both immediate liquidity concerns and broader concerns about the platform's security.  Since Drift is the most significant decentralized perpetual futures exchange in the Solana ecosystem, the compromise has implications that go beyond a single protocol, and it raises new concerns about systemic risk, adversarial persistence, and the resilience of decentralized trading infrastructures in the face of sustained, state-aligned threat activities.  A Drift Protocol internal assessment further suggests that the breach was the culmination of a deliberate and six-month intrusion campaign. The activity was attributed with moderate confidence to a North Korea-aligned threat cluster identified as UNC4736.  There are numerous aliases for this actor, including AppleJeus, Citrine Sleet, Golden Chollima and Gleaming Pisces. This group has a long history of financial motivated intrusions within the cryptocurrency threat landscape, as evidenced by its track record of financial motivations. It is noteworthy that the group's past activity has been associated with high-impact incidents such as the X_TRADER and 3CX supply chain compromises of 2023 and the Radiant Capital breach of late 2024, both of which resulted in $53 million losses.  As a consequence of Drift's analysis, transactional continuity and operational continuity can be demonstrated by observing the preparatory fund movements that were associated with the exploit that were traceable to earlier attacks.  Additionally, the social engineering framework demonstrated measurable overlap with previously documented DPRK-linked campaigns in terms of persona construction and engagement tactics. This attribution is supported by independent threat intelligence reports. CrowdStrike's January 2026 assessment identifies Golden Chollima as an offshoot of the DPRK cyber apparatus that performs sustained cryptocurrency theft operations against smaller fintech companies throughout North America, Europe, and parts of Asia as part of its ongoing cyber warfare efforts.  Based on the group's methodology, it appears that the group is pursuing consistent revenue streams through repeated, lower-profile compromises in favor of singular, high-profile events. In line with the regime’s broader strategic imperatives, cyber-enabled financial theft is seen as an effective means of balancing economic constraints and supporting long-term military and technological objectives.  As observed, UNC4736 engages in social engineering with precision, as well as post-compromise technical depth. A documented case from late 2024 illustrates how the group utilized a fabricated recruitment campaign to distribute malicious Python packages, establishing a foothold in a fintech environment within Europe. A lateral movement into cloud infrastructure enabled access to identity and access management configurations, which enabled diversion of digital assets to adversary-controlled wallets as a result of this access. It is becoming increasingly apparent, within this context, that the Drift incident is not merely an isolated exploit, but rather an intelligent intelligence operation that was conducted with patience and strategic intent.  In collaboration with law enforcement agencies and forensic specialists, the platform is reconstructing the intrusion timeline, and initial indications suggest an organized progression from reconnaissance and access acquisition to staged execution and asset extraction.  An examination of the larger operational ecosystem underpinning such campaigns reveals a highly structured, multinational workforce model designed to sustain long-term access and revenue generation. A distributed network of technical proficient individuals is employed by the program, many of whom operate in jurisdictions such as China and Russia.  Through company-issued systems hosted in geographically dispersed laptop farms, including within the United States, employees are remote interacting with corporate environments. It is supported by an intermediary layer of facilitators who coordinate logistical tasks, which include handling devices, processing payroll, and establishing identity credentials, which are often orchestrated through shell entities aimed at obscuring attribution and bypassing regulatory scrutiny.  In itself, the recruitment and placement pipeline exhibits a degree of operational maturity which is commonly associated with legitimate global hiring ecosystems. As part of the initial recruitment process, dedicated recruiters identify potential candidates, followed by a structured onboarding process in which curated identities are assigned and refined.  Facilitators are responsible for managing professional profiles, directing summary development, and conducting targeted interview coaching, ensuring alignment with Western employers' expectations. The use of enhanced verification mechanisms involves the introduction of additional collaborators in order to satisfy compliance checks, thereby effectively bridging the gap between fabricated personas and real-world hiring requirements. This model relies on cryptocurrency for the financial backbone, allowing wages to be systematically repatriated while minimizing exposure to international sanctions.  Furthermore, threat intelligence reports indicate that this workforce is deliberately transient by design. Employees frequently change roles, identities, and digital accounts, maintaining a fluid presence that complicates detection and attribution.  By reducing exposure risk for a long period, constant churn enables continuous infiltration across multiple organizations simultaneously and reduces the risk of long-term exposure. A recent study indicates that the recruitment base has been expanded beyond traditional boundaries, with individuals from Iran, Syria, Lebanon, and Saudi Arabia actively participating in the program.  A number of documented examples demonstrate the effectiveness of the model in advancing candidates from these regions through employment processes with U.S.-based employers. Within this framework, there has been an important development in the use of legitimate professional networking platforms to recruit auxiliary participants individuals who are responsible for performing real-time interactions such as technical interviews in under assumed identities.  The participants, often trained and evaluated through recording sessions, serve as proxies for obtaining employment positions based upon fabricated Western personas. Such access can be used for a variety of intelligence purposes once embedded, as well as financial extraction.  While monetary gains remain the primary motivation, the intentional targeting of sectors such as the defense contracting industry, financial services, and cryptocurrency infrastructure suggests a convergence of economic and strategic objectives. In the aggregate, these developments reveal a highly sophisticated, multi-layered strategy that extends far beyond conventional cybercrime, blurring the distinction between the infiltration of workers, espionage activities, and financial operations carried out by the state.  As a whole, the incident illustrates a convergence in advanced intrusion capabilities and increasingly institutionalized support architecture that goes beyond conventional definitions of cybercrime. A well-crafted exploit is not the only thing that emerged from the Drift breach, but a deeply embedded operational system that integrates financial theft with identity theft and worker infiltration.  Considering how large the assets were exfiltrated, along with the precision with which transactions were staged and laundered, one can conclude that these campaigns were neither isolated nor opportunistic, but rather were part of an ongoing and adaptive model operating across jurisdictions, platforms, and regulatory environments. As a result of the attribution indicators viewed together with historical activity, a continuity of intent and methodology has been identified that is consistent with long-observed DPRK-linked activity. In light of the interplay between on-chain movement patterns, infrastructure reuse, and human manipulation, a hybrid threat approach is being developed, which combines technical compromise with social engineering and operational deception.  Through this dual-layered methodology, threat actors can not only amp up the effectiveness of individual attacks, but also enhance their persistence, making it possible for them to reconstitute revenue streams and access after partial disruptions. This instance highlights the inherent tension between innovation and security within rapidly evolving financial architectures, as well as its systemic implications for the broader digital asset ecosystem.  As a result, critical questions emerge regarding trust assumptions within decentralized environments, the effectiveness of monitoring mechanisms for complex transaction flows, and the readiness of platforms to counter adversaries who operate both strategically and with state-level resources. In the coming months and years, the Drift incident is likely to be viewed less as a single breach and more as an example of state-administered cyber-financial operations maturing.  Throughout the digital domain, economic objectives, geopolitical strategies, and technical execution are increasingly converged. This is creating a threat landscape that challenges traditional defensive models and requires both industry and government stakeholders to respond more intelligently and integrated.  Accordingly, the Drift incident illustrates the emergence of highly sophisticated intrusion capabilities and an increasingly formalized operational ecosystem that is well beyond the traditional frameworks used by cybercriminals. In addition to the exploitation of a technically complex exploit, the breach reveals the existence of a larger, deeply embedded apparatus that, in its unified and scalable form, systematically combine financial extraction, identity manipulation, and workforce infiltration. With such a large amount of asset exfiltration combined with calculated sequencing of fund movements and obfuscation, it is evident that such operations are deliberate, repeatable, and designed to operate across diverse regulatory and technological environments. Upon contextualization with prior activity, the attribution signals suggest a consistent alignment of intent and execution, consistent with long-documented DPRK-linked campaigns.  As a consequence of the correlation between on-chain behavioral patterns, reuse of operational infrastructure, and coordinated human-centric tactics, it is apparent that a hybrid threat model is being developed in which technical compromise and controlled deception are inseparable.  As a result of this layered approach, operational success rates are increased as well as resilience is achieved, enabling threat actors to re-establish footholds and maintain financial output even in the event of partial exposure or disruption. This has material implications for the wider ecosystem of digital assets.  A prominent decentralized derivatives platform has been compromised, bringing into sharp relief the inherent trade-off between rapid innovation in financial markets and robust security measures. As a result, decentralized systems are once again in the spotlight, causing us to examine the role trust plays within them, the effectiveness of existing transaction monitoring frameworks, and the overall readiness of platforms to combat adversaries who have strategic foresight and state backing.  In time, as investigations progress and details of attribution become clearer, the breach may serve as a useful historical reference point for understanding how state-aligned cyber-financial operations have changed over time.  Economic imperatives, geopolitical objectives, and technical sophistication are now convergent within the cyber domain, which is redefining threat paradigms and reinforcing the need for coordinated, intelligence-driven defense strategies both within the public and private sectors.