The cross-chain bridge of Web3 platform The Sandbox was exploited on August 21, 2026, after an attacker abused a vulnerability in the approveAndCall function and triggered the large-scale minting of unbacked SAND tokens on Base. According to onchain data, 329 trillion phantom SAND were minted on Base, theoretically worth $49 billion at market prices — an amount dozens of times larger than the token's actual supply [[10]].
Key fact: Despite the enormous nominal amount, structural limitations of the bridge prevented the attacker from extracting the entire sum. The actual financial damage is measured in billions rather than tens of billions of dollars, although the psychological impact on the market was substantial.
PeckShield was the first to detect abnormal activity, reporting 14.9 billion unbacked SAND appearing across Base and BNB Chain [[5]]. According to deeper analysis, the attacker exploited the approveAndCall function within the LayerZero-based infrastructure connecting SAND to Base and BNB Smart Chain [[15]]. The function allows token spending approval and arbitrary logic execution to occur atomically within a single transaction — a design that can become a vector for hijacking minting privileges if validation is implemented incorrectly.
The attacker seized minting authority and used it to create 329 trillion unbacked SAND on Base. However, the cross-chain bridge does not support unlimited withdrawals: structural limits on outgoing liquidity and batch operations made it impossible to immediately move the entire supply to Ethereum mainnet or exchanges [[10]]. As a result, the amount that could actually be extracted was substantially smaller than the headline figure.
Public reports contain two very different figures, and both can technically be correct because they describe different aspects of the incident. The 329 trillion figure represents the amount the attacker was technically able to mint on Base after obtaining minting authority [[10]]. The 14.9 billion figure represents the volume that actually moved through the bridge or was identified by PeckShield as unbacked tokens effectively issued into circulation [[5]].
The discrepancy exists because minting authority does not equal liquid supply. Most of the newly minted tokens remained in the attacker's wallet without a realistic way to monetize them through legitimate DEX liquidity pools on Base, which were quickly frozen or had their liquidity withdrawn. This is a rare example in which the theoretical "size of the hack" vastly exceeds the actual financial damage — precisely because bridge-level rate limits and whitelist controls restricted how much value could leave the compromised environment.
Lesson for developers: Even when minting authority is compromised, bridge-level safeguards such as rate limits, outgoing caps and whitelists can reduce actual losses by orders of magnitude. Defense in depth worked.
The Sandbox team followed the standard playbook for a cross-chain security incident: bridges on Base and BNB Chain were immediately disabled to isolate the unbacked tokens and prevent their further spread [[17]]. Users were strongly advised not to trade SAND on Base or BNB Smart Chain while liquidity on those networks remained at risk and until the affected pools had undergone a full review [[9]].
The project's official position was cautiously worded: the incident was characterized as "minimal" in terms of actual financial damage, while potentially serious from a reputational perspective [[9]]. The Sandbox launched an investigation with security auditors and onchain analysts to determine the exact approveAndCall exploitation vector and establish whether the primary SAND contract on Ethereum had been affected.
The incident triggered a rapid response from South Korean exchanges. Upbit and Bithumb suspended SAND deposits and withdrawals under the country's Virtual Asset User Protection Act [[8]]. Upbit suspended operations not only for SAND on Base and BNB Chain but also for the Ethereum version as a precautionary measure until the project team could formally confirm the security of the main contract [[8]].
This represented an unusually conservative response. Korean exchanges have historically focused restrictions on the networks directly affected by an incident. In this case, the logic was broader: any uncertainty surrounding the integrity of the token justified temporarily restricting all supported versions in order to prevent cross-chain arbitrage or the laundering of unbacked tokens through legitimate liquidity.
News of the exploit triggered a sharp sell-off in SAND across major exchanges. The token fell by double-digit percentages in the first hours following PeckShield's report. The primary concern for the market was not simply the number of newly minted tokens but information asymmetry: traders could not immediately distinguish between "backed" SAND in Ethereum liquidity pools and "phantom" SAND on Base, encouraging liquidity providers and traders to reduce exposure across SAND markets.
Coinbase and other major custodians had not suspended Ethereum-based SAND operations at the time, but continued monitoring onchain flows. A key requirement for restoring confidence will be the publication of a security assessment and a clear separation between legitimate and compromised liquidity pools.
SAND users should take three precautions: (1) avoid interacting with SAND contracts on Base and BNB Smart Chain until The Sandbox issues an official all-clear; (2) review wallet approvals through Revoke.cash and revoke permissions granted to suspicious bridge contracts; and (3) avoid moving SAND between networks until the investigation is complete, as even legitimate tokens may temporarily be classified as high-risk by exchanges and compliance systems.
Practical advice: If you hold SAND on an exchange, monitor its deposit and withdrawal status. Upbit and Bithumb have already suspended operations, and other platforms may introduce similar restrictions as additional information becomes available.
Vulnerabilities involving approveAndCall are not new to DeFi. Similar patterns have previously created risks for ERC-20 integrations in which an attacker could combine token approval with callback logic executed by the receiving contract. In SAND's case, the risk was amplified by the cross-chain architecture, where approveAndCall formed part of the mechanism used to coordinate token operations between networks.
The industry has increasingly moved toward alternative authorization and cross-chain execution patterns, including Permit2-style permissions, intent-based cross-chain standards and native mint-and-burn architectures designed to reduce reliance on broad token approvals. The Sandbox incident could accelerate scrutiny of approveAndCall implementations across bridge infrastructure.
LayerZero, as the infrastructure provider underlying the bridge integration, is likely to face increased scrutiny. Although the vulnerability appears to have existed at the level of SAND's integration rather than in LayerZero's core protocol, reputational effects can spread across the broader ecosystem. Other projects using similar integration patterns may consequently conduct additional audits of their own bridge implementations.
For the market, the incident is another reminder that cross-chain bridges remain one of the most sensitive attack surfaces in DeFi infrastructure. Earlier incidents involving Ronin, Wormhole, Nomad and Harmony demonstrated the enormous value concentrated in bridges. The difference in more recent designs is the increasing use of rate limits and other containment mechanisms capable of restricting losses even after part of the system has been compromised.
The incident highlights four important lessons. First, the nominal size of an exploit and the actual financial loss are different metrics, and defense-in-depth mechanisms can dramatically reduce the latter. Second, cross-chain functions such as approveAndCall require dedicated security reviews regardless of the security of the underlying protocol. Third, exchanges and regulators increasingly respond preemptively, restricting assets before investigations are complete. Fourth, a project's communication during the first hours of an incident can significantly influence the scale of market panic.
Strategic takeaway: Cross-chain infrastructure is no longer a "nice-to-have" extension of a protocol — it is a critical attack surface. Projects that fail to subject bridge integrations to rigorous security reviews risk both user liquidity and their reputation.
The ultimate consequences of the incident will depend on the investigation. If The Sandbox confirms that the primary Ethereum SAND contract remained unaffected, the ecosystem may be able to isolate the compromised supply and restore normal operations relatively quickly. If minting authority or related privileges were found to extend to additional networks, a broader redesign of the token's cross-chain architecture could become necessary.
“A bridge is not simply a pipe between networks. It is infrastructure entrusted with other people's assets, and every component of it must be designed with that responsibility in mind.”
— On the security risks of cross-chain infrastructure
