On Oct. 24, a single Ethereum transaction borrowed about 1.94 million USDT from Aave, bought X tokens and froze them through two new helper contracts. Each helper locked 19,000.000001 X, just above the incumbent senator's stake, and took two of the three seats on X100's Senate, the body that controls its treasury. The helpers then created, approved and executed a proposal in the same transaction.

The proposal made the treasury sell 400,000 X into a pool the attacker had created moments earlier with about $100 of liquidity, so the treasury received just $35.68. The attacker sold X into the real market, repaid the flash loan and kept about $4.75 million in USDT. A second actor reused the same path for about $96,000, bringing realized USDT losses to roughly $4.85 million, and a third pulled about 550,000 X from remaining treasury liquidity. The X token fell more than 96%. Some early reports put the loss near $6 million, but on-chain reconciliations from Blockaid, Bitquery and others support about $4.85 million in USDT, excluding the value of extracted X.

The flaw was not the size of the loan but the absence of any gap between acquiring influence and using it. Senate eligibility used the amount currently frozen, with no minimum stake age, and the voting period limited how late a proposal could run but not how early it could execute.

Standard controls would have broken the attack: voting power fixed by a snapshot taken before a proposal opens, a minimum stake age for eligibility, and a timelock between approval and execution. Any protocol whose governance can move treasury funds should treat it as an access control system, where authority that can be assembled and used in one transaction is effectively no control at all.