Mentioned in This Article
1. A One-Dollar Test Shows How Fast Bots Sweep Weak Keys
Austin Griffith of the Ethereum Foundation sent about $1 (0.0004 ETH) to an Ethereum wallet whose private key was simply the number 6969, and bots tried to take it within seconds. Griffith described the experiment on October 21 on a podcast he co-hosts. The result shows that an Ethereum weak private key, one that is easy to guess, offers essentially no protection.
Bots were on the wallet almost at once. Griffith said that three transactions were pending in the mempool, each trying to steal the money, as soon as the dollar arrived. His words were: "As soon as the dollar landed, there were three transactions in the mempool pending trying to steal it."
The mempool is the waiting area where Ethereum transactions sit after they are broadcast and before a block includes them. Anyone running a node can watch it, which is how automated sweepers spot a freshly funded address and race to move its balance before the owner can react.
Such bots are not sophisticated in concept. A sweeper holds a list of known weak keys, derives the address for each, and monitors those addresses continuously. When a balance appears, it signs a transaction with the known key and broadcasts it with a fee high enough to be included quickly. When several operators run the same strategy, they compete, which explains why more than one pending transaction appeared for a single dollar.
The scale of the money involved is irrelevant to the machines. Because the monitoring is automated and cheap, a balance worth one dollar is treated the same way as a balance worth much more, and the speed of the response is measured in seconds, not in hours or days.
2. What Is a Private Key and Why Does a Weak One Matter?
A private key is the secret number that controls an Ethereum wallet, and whoever knows or guesses it can move the funds. The wallet's public address is derived from that number, but the derivation cannot be run backward in any practical way. Security therefore rests entirely on how hard the number is to guess.
A properly generated key is drawn from an astronomically large range, so searching for it by chance is hopeless. The protection disappears if a person chooses the number by hand. A key like 6969 sits at the very start of the range, and a bot that counts upward from 1 reaches it almost instantly. Funds sent to such an address can be assumed to belong to whoever sweeps them first.
This is why wallet software generates keys from random data and asks users to back up a recovery phrase instead of inventing their own number. The one-dollar test was designed to make that principle visible by showing what happens in practice rather than in theory.
It also helps to separate two things that are often confused: a wallet that is hacked and a wallet that was never secure. In this case nothing was broken in Ethereum itself. The network processed valid transactions signed with a valid key, exactly as designed. The weakness lay entirely in the choice of secret, and no protocol feature can rescue funds once the key is public knowledge or trivially guessed.
3. Why Did Griffith Run the Test?
Griffith ran the test after a security failure at eMudhra, an Indian certificate authority. A certificate authority is an organization that issues the digital certificates browsers use to confirm that a website is genuine, and its signing keys are meant to be among the most closely guarded secrets in the system.
A disclosure site set up by Dennis Jackson of Mozilla reported that the keys behind eight eMudhra intermediate certificates, created in April 2025, had been compromised. Researcher Scott Helme wrote that each of those keys was a number below 256, and one of them was 4. Keys that small can be recovered by trying every possibility in a few moments on ordinary hardware.
The response from browser makers was swift. Firefox and Chrome blocked the certificates on October 17, the day the problem was reported. Microsoft followed on October 20. On October 22, after the podcast episode was recorded, eMudhra said it would exit the global SSL/TLS certificate business.
The incident supplied Griffith with an obvious question: if a certificate key could be a tiny number, how quickly would the Ethereum ecosystem's bots punish an equally tiny wallet key? The experiment answered it.
The same reasoning explains why the certificate failure drew such a fast reaction. Browsers trust certificates issued by authorities on an approved list, and an intermediate certificate lets its holder vouch for other websites. If someone can derive the key behind such a certificate, they could in principle create certificates for sites they do not control, which is why blocking the affected certificates was treated as urgent.
4. What Do the On-Chain Records Show?
The blockchain data supports Griffith's account. Records show that 0.0004 ETH, worth about $1, reached the address for key 6969 at 20:35 UTC on October 21. A transaction signed with the same key moved the money out in the next block, 12 seconds later, and most of the deposit went to network fees rather than to the thief.
Transactions from three other addresses reached the wallet just after the deposit, and two of them failed. That pattern matches what Griffith described: several competing bots reacted to the same funding event, and only one could win. The deposit itself came from a wallet named slop.atg.eth, which had been funded by austingriffith.eth.
The fact that most of the stolen amount was consumed by fees shows how marginal the prize was. Even so, the bots acted, because their cost of watching is tiny and the payoff, however small, is automatic.
In the account the records support, the sequence is short. A deposit arrives, competing transactions appear, one is included in the next block, and the balance is gone. There is no step at which the original owner could have intervened, because Ethereum blocks arrive roughly every 12 seconds and the sweeper acted in the very first one available.
5. Is Any Number Below 256 Still Safe to Use?
No, because every such key has already been used. Griffith pointed out that there is no number under 256 that lacks a nonce, the counter of transactions an address has sent. His phrasing was: "there's not a number less than 256 that doesn't already have a nonce."
A check of the addresses for keys 1 through 255 on Ethereum mainnet bears this out. Every one of them has sent transactions. A nonzero nonce means that at some point someone funded the address and the funds were swept out, which fits the pattern of bots continually walking the lowest part of the key space.
The practical meaning is that low-number keys are not merely guessable in principle. They are actively and repeatedly harvested, and any value sent to them will be gone almost immediately.
6. How Long Can a Guessable Key Survive?
Griffith then turned to longer keys to see where the bots stop looking. Before the results came in, he guessed that bots would stop finding keys somewhere "between eight and 12" digits. Later in the episode he said that 10-digit keys had been drained "right away," while 12-digit keys had "not yet" been drained.
He also said he had funded a 14-digit key and a 16-digit key, "looking for someone to steal my dollar." He did not name those keys, and it is not possible to verify independently which ones he used. His remarks are therefore best read as his own account of an informal experiment, not as a controlled study.
On-chain records add partial detail. The slop.atg.eth wallet funded four more addresses about 14 minutes after the first deposit. One of them was emptied within 12 seconds, while three still held their funds as of October 23. The records do not show which key length corresponds to which address, so they cannot confirm which of the longer keys, if any, have been found.
The comparison between 10-digit and 12-digit keys is nonetheless instructive. Each extra digit multiplies the number of candidates a bot must test by ten, so a 12-digit key has a hundred times as many possibilities as a 10-digit one. The experiment is an attempt to locate the point where the cost of searching outweighs a one-dollar reward.
7. What Comes Next for the Experiment?
The hosts said they would check the following week whether the 14- and 16-digit keys had been drained. That follow-up will indicate whether bot operators are willing to search that far for tiny balances, or whether keys of that length are already out of reach for the economics of dust-sized theft.
Whatever the outcome, the lesson for ordinary users does not change. Wallet keys should come from software that generates them randomly and should never be chosen from memorable numbers or phrases. For the broader industry, the eMudhra case shows that the same principle applies to the infrastructure of the web: a signing key that can be guessed provides no assurance, whether it guards a single dollar or the identity of websites.
Developers building wallets and tools can draw a second conclusion. Any feature that lets a user type in a short or memorable secret, even for testing, risks leaving real funds exposed. Test networks exist so that experiments with throwaway keys carry no financial consequence, and mainnet addresses derived from simple numbers should be treated as public property from the moment they are created.