A stale-cache bug in Aptos's Move VM gave researchers a near-90% success rate at breaking the chain's core security guarantees, with attack costs of just hundreds of dollars.
Back
Aptos bug put $70 billion in crypto at risk, researchers find
A stale-cache bug in Aptos's Move VM gave researchers a near-90% success rate at breaking the chain's core security guarantees, with attack costs of just hundreds of dollars.
A $3,000 server was enough for security researchers to simulate an attack on the Aptos blockchain that could have put as much as $70 billion in crypto infrastructure at risk, according to findings published Friday.
"It ran as claimed, and the exploit made sense," Mudit Gupta, chief technology officer at Polygon, told CoinDesk after independently reviewing the proof-of-concept materials. "It required a few conditions to be met, which it seems like they did on the mainnet."
Researchers at Hexens, a blockchain security firm, identified a stale-cache bug leading to a type-confusion vulnerability in the Aptos Move virtual machine, the execution environment that processes smart contracts on the chain. The team ran the exploit path roughly 20 times in a simulated environment and succeeded 17 or 18 times — a near-90% success rate — using a server setup that cost about $3,000 to simulate roughly one-third of the validator network. The attack required no insider access or special permissions.
The vulnerability was reported through SEAL911 emergency channels on Feb. 25, and Aptos deployed a patch to mainnet within hours. No funds were lost. But the disclosure shows how a single undetected bug in a layer-1 blockchain's execution layer could cascade through bridges, stablecoin issuers and centralized exchanges, turning a contained protocol flaw into a market-wide crisis.
How the bug worked
The sensitivity of this class of bug comes down to how the Move language handles authority. Protocol permissions in Move — including the right to mint a stablecoin, control a bridge, or administer a lending market — are stored directly as onchain resources. If those resources are compromised, the damage extends to everything that trusts them.
Hexens' researchers compared the bug to a flaw on an Ethereum-style chain that would allow attacker-controlled code to write into storage belonging to other contracts, bypassing the type-system guarantees that Move was specifically designed to uphold.
Grego AI, which independently verified Hexens' proof-of-concept, calculated that approximately $250 million in Aptos-native total value locked was directly at risk based on the near-90% success rate, separate from broader cross-chain exposure.
The $70 billion systemic risk
Hexens assessed that the broader first-order systemic risk was approximately $70 billion — a figure that includes value accessible through bridges, cross-chain messaging systems, stablecoin administration flows and centralized exchanges. Grego AI noted the exploit could also be used to steal protocol capabilities held by LayerZero, Wormhole and USDC's Cross-Chain Transfer Protocol.
"If malicious actors had access to this bug, they would have been able to take all the TVL that they wanted," Justus Hanna, chief executive officer at Grego AI, said.
The $70 billion estimate is based on minting a large amount of USDC and using Circle's CCTP to move it across chains. In practice, Circle could halt USDC transfers, though the stablecoin issuer has faced scrutiny after saying it does not freeze assets without legal authorization. Even so, the researchers demonstrated access to master-minter roles, bridge signer capabilities and protocol accounting state — the kinds of authority that sit at the top of cross-chain systems.
The dominant vector into the broader surface runs through centralized exchanges, specifically the Aptos bridge pathways that connect onchain activity to exchange deposit crediting.
Response and disclosure
The same day Hexens filed its report, a SEAL911 emergency warroom was opened to coordinate the response. The vendor was notified hours later, and four major downstream projects were alerted that afternoon, each receiving local-runnable proof-of-concept material and analysis of relevant authority patterns.
A public pull request reflecting the patch became available on Feb. 27. Aptos stated that a private-validator patch had been deployed before the public commit. An Aptos spokesperson told CoinDesk that "a fix was developed, tested, and deployed to mainnet within hours of discovery. No users or funds were impacted at any point." The spokesperson also disputed the practical exploitability of the bug, saying the company's analysis determined it would have "extremely low exploitability in real world conditions."
Hexens said it has not received a technical rebuttal or evidence-based argument disputing the demonstrated impact classes. The firm said the main concern relayed back to the researchers involved the probabilistic aspects of the exploit — precisely what the team's calibration work was designed to address.
If an attacker had found and exploited the bug, it could have dwarfed the $1.5 billion stolen in the Bybit hack last year. In June, Zcash fell 38% after developers revealed a critical bug that had gone undetected in its privacy pool for four years, one that could have allowed an attacker to print unlimited counterfeit tokens. The Aptos disclosure adds to a growing list of near-misses that test the industry's ability to detect and patch critical flaws before they are exploited.
This article is for informational purposes only and does not constitute investment advice.
