Dormant Truebit Contract Drained of $26.2 Million Via Five-Year-Old Overflow Bug

On January 8, 2026, an attacker emptied Truebit's aging purchase contract of 8,535 ETH (roughly $26.2 million) in a single transaction, capping off nearly five years of dormancy for a contract that had sat unaudited and unverified on Ethereum since 2021.

01A warning nobody acted on

The exploited contract had been public knowledge as a risk for years. Its bytecode was never verified on Etherscan, no audit had ever been published, and its bonding-curve design was called out by pseudonymous researcher Banteg as far back as 2021, who labeled it a potential "rug zone." Rekt itself had written about the project's turbulent launch at the time. None of that scrutiny translated into a fix, and the contract simply sat untouched until someone decided to test it.

That someone struck on January 8, becoming, as The Block reported, the first significant DeFi loss of the new year. The attacker minted more than 240 million TRU tokens, immediately burned them for real ETH, and repeated the cycle until the protocol's reserves were empty and 8,535 ETH had moved into Tornado Cash. Observers later noted a darkly fitting detail: the function the attacker invoked to carry out the theft was literally named "Attack".

The TRU token did not just fall — it was effectively erased. Price data from CoinGecko shows it going from $0.16 to a fraction of a cent within hours, undercutting the project's own tagline, still displayed on its Twitter/X account: "Don't just trust, verify." PeckShield subsequently linked the attacking wallet to an earlier, much smaller drain of the Sparkle protocol for 5 ETH twelve days before — suggesting this was not a one-off opportunist but someone working methodically through a list of neglected contracts.

02How the exploit unfolded, minute by minute

Monitoring firm Cyvers was first to flag the activity, catching the 8,535 ETH leaving Truebit's Purchase contract in one sweep. Within three minutes, researcher TheDeFiDan had already spotted the "Attack"-labeled function that would become the incident's grim signature.

Independent analysts moved quickly to characterize the bug. William Li offered early commentary noting the target was "a very old contract deployed ~5 years ago," and warned that such legacy contracts were becoming an increasingly popular target category. Separately, BlockScope traced the attacker's funding trail back to November, pointing to weeks of advance preparation routed through Rhino.fi. The most consequential thread, though, came from PeckShield, which connected the exploiter's wallet to the Sparkle protocol incident less than two weeks earlier — evidence of a repeat operator rather than a first-timer.

Truebit itself stayed silent for roughly two hours before issuing a statement on X: "Today, we became aware of a security incident involving one or more malicious actors. The affected smart contract is 0x764C64b2A09b09Acb100B80d8c505Aa6a0302EF2 and we strongly advise the public not to interact with this contract until further notice." The statement gave no technical explanation and made no mention of the roughly $26.2 million already gone, promising only cooperation with law enforcement.

A second, smaller actor also capitalized on the chaos, reportedly boasting in a group chat about pulling in roughly $250,000 in copycat profits once the exploit became visible on-chain. Meanwhile TRU's price kept collapsing — from $0.16 down to roughly $0.000000018 — as liquidity pools drained faster than holders could exit, leaving anyone still in the token with a near-total loss. It marked the first major DeFi hack of 2026, occurring before the month was even half over.

03The mechanism: an unguarded addition in a five-year-old formula

Truebit's contract, deployed in 2021, operated on a straightforward premise: users could mint TRU with ETH and later burn TRU to reclaim ETH. Minting followed a bonding curve, growing more expensive as circulating supply rose. Redemptions worked on a different rule entirely — the protocol would buy back tokens at only 12.5% of the highest price ever paid to mint them, the exact mechanic Banteg had flagged in 2021 during the project's rocky debut.

The flaw sat inside getPurchasePrice(uint256 amount), the function responsible for calculating how much ETH a given mint order should cost. It behaved correctly for realistic input sizes, but broke down when fed values far beyond any plausible token supply. The root cause, according to analysis from 0xkaka, was the contract's compiler version, 0.5.3 — old enough to predate Solidity's automatic overflow protection, meaning addition operations could silently wrap around without reverting. The contract used SafeMath to guard its multiplication and division operations, but one addition inside the pricing formula was left unprotected.

Feeding that addition an extreme value caused it to overflow and wrap around to a number near zero. As a result, the cost to mint 240,442,509 TRU tokens rounded down to almost nothing.

From there, the attack contract executed a repeatable loop, visible in the transaction's event log: call getPurchasePrice() with an oversized amount to get a near-zero quote, mint billions of TRU tokens for that negligible cost, approve and forward those tokens to the Purchase contract, burn them for real ETH at the 12.5% buyback rate, then repeat with an even larger figure. The whole sequence ran five times within one transaction, with no price movement between steps since everything executed atomically. Nothing in the contract — no supply ceiling, no per-transaction cap — was in place to interrupt the cycle, and it continued until 8,535 ETH had passed from the protocol's reserves into the attacker's control. As AstraSecAI put it, "a reminder that one missed check is all it takes."

Because the contract's source was never verified on Etherscan — only compiled bytecode was public — finding this bug required decompiling the contract first, implying the attacker had put in real reverse-engineering effort before striking.

04Tracing the money

BlockScope's on-chain analysis placed the attacker's initial funding in November 2025, well ahead of the exploit itself. The primary attacking address is 0x6c8ec8f14be7c01672d31cfa5f2cefeab2562b50, which received its seed funds via the Across Protocol bridge on December 6, 2025, in transaction 0xf173f0ed341d3106c1f2eda4704a5e5e9e12c8cf896eb525948624841e7d7ad.

The targeted contract — Truebit's Purchase contract at 0x764C64b2A09b09Acb100B80d8c505Aa6a0302EF2 — had been deployed roughly five years prior, its bytecode never verified and never audited. The attacker's dedicated exploit contract sits at 0x1de399967b206e446b4e9aeeb3cb0a0991bf11b8. According to Extractor by Hacken, this contract was deployed in the very same block as the exploit itself, submitted through a private mempool — deployment occupied position 3 in the block, execution position 4 — leaving no window for front-runners and ensuring clean, uninterrupted sequencing.

The primary attack transaction, 0xcd4755645595094a8ab984d0db7e3b4aabde72a5c87c4f176a030629c47fb014, contained all five mint-and-burn cycles, extracting the full 8,535 ETH and leaving the Purchase contract with only about 16 ETH remaining.

Funds began moving almost immediately after the exploit. From the original attacking wallet (0x6C8EC8f14bE7C01672d31CFa5f2CEfeAB2562b50), proceeds passed through a chain of intermediary addresses: Laundering Wallet 1 at 0x273589ca3713e7becf42069f9fb3f0c164ce850a, Laundering Wallet 2 at 0x3b58192943ee6f9ae92d54dd1ef378cfd519862a, Laundering Wallet 3 at 0x62afdd1bd84f6b152572404be90679ae58eb4862, and Laundering Wallet 4 at 0xD12f6E0fa7FBF4e3A1c7996E3F0Dd26AB9031a60. On-chain tracker Lookonchain confirmed on January 10 that the full 8,535 ETH had by then been deposited into Tornado Cash, via wallet 0xD841C52B68c5dB133078ABa039bd9EAF19b0b135, completing the laundering process.

Once the exploit was visible on-chain, MEV bots and copycat attackers moved in to capture whatever remained. A second attacker, using wallet 0xc0454E545a7A715c6D3627f77bEd376a05182FBc and with no apparent link to the first attack, walked away with approximately $253,000 in opportunistic profit.

The most notable link, however, remained PeckShield's discovery that the primary attacking wallet had drained the Sparkle protocol for about 5 ETH just twelve days earlier — a roughly 1,700x jump in yield between the two incidents in under two weeks, consistent with a repeat operator methodically working down a list of vulnerable, unmonitored contracts rather than someone who got lucky once.

05Part of a pattern: DeFi's abandoned-code problem

Truebit's loss adds to a lengthening list of 2025-2026 incidents traced back to old, unmaintained contracts. Balancer lost $128 million on November 3 to rounding errors in Composable Stable Pools that had gone unexamined for five years, spread across multiple chains. Yearn's yETH lost $9 million on November 30 when an unmaintained stableswap pool with unusual math allowed unlimited token minting. Abracadabra lost $1.8 million on October 4 through CauldronV4 contracts that had been labeled "deprecated" but were never actually disabled. Aevo (formerly Ribbon Finance) lost $2.7 million in December to a proxy admin vulnerability in old vaults, and Rari Capital lost roughly $2 million in December via a multisig takeover of a protocol that had already shut down operations.

Former Yearn developer stormblessed summed up the recurring theme: "Another victim of legacy code? It's going to keep happening," adding that "teams should assume old code is being actively looked at and either deprecate/sunset or reaudit."

Anthropic's own research adds weight to that warning: when its AI agents were tested against 405 previously exploited smart contracts, they autonomously produced $4.5 million in successful exploits, including against contracts deployed after the models' training cutoff — evidence that the technical bar for rediscovering this class of bug keeps falling.

Truebit's own path to this point began with a chaotic 2021 launch marked by a stealth deployment, confused early investors, and Banteg's contemporaneous warning about the tokenomics' rug-zone risk. None of that led to a published audit or verified source code in the five years since. As of this writing, Truebit's most recent update, posted January 9, says only that the team has "engaged additional resources to strengthen tracing and recovery" — with no compensation plan or recovery timeline disclosed.

The attacker who emptied Truebit's contract had previously taken 5 ETH from Sparkle and escalated to 8,535 ETH within about two weeks. Whether that pattern continues against other dormant contracts elsewhere on Ethereum remains an open question.