464 ETH. Four deposit wallets. One intermediate hop. And roughly 700 ETH still parked in the source address, untouched.

That is the entire public footprint of the CryptoBilis breach. No press release with a loss figure. No timeline. No confirmation of whether user funds were touched. Just a transaction graph that anyone with an RPC endpoint and enough patience can reconstruct.
Here is the part most coverage skips: 464 does not divide by 100. Tornado Cash runs fixed-denomination pools โ 0.1, 1, 10, and 100 ETH. Four hundred is clean. The trailing 64 is not. Someone deliberately mixed pool sizes, or routed value through an extra hop, or both. That residue is not noise. It is a signature. Narrative is the new liquidity โ and this attacker just minted a very specific one.
To read the event, you need the Tornado Cash backstory, because the tool's legal status flipped twice while most people were watching price charts.
August 2022: OFAC sanctions Tornado Cash, adding the protocol to the SDN list. Every US person touching it is suddenly exposed. Developers get charged โ Alexey Pertsev convicted in the Netherlands, Roman Storm facing trial in the US. November 2024: the Fifth Circuit rules OFAC overstepped its statutory authority. March 2025: the Treasury delists the protocol.
So when 464 ETH walks into Tornado Cash today, it is not, strictly, a sanctions violation. It is a compliance flag. Chain-analytics firms still score those interactions as high risk. Exchanges still freeze deposits with tainted provenance. The tool is legal and radioactive at the same time โ a distinction that matters enormously and that almost no headline captured.
Let me walk the graph the way I'd walk it in a tracing engagement. CryptoBilis gets drained. Funds move to an attacker-controlled wallet. Then an intermediate address โ a hop that exists for exactly one reason: to break the naive "funded by" heuristic that most off-the-shelf clusterers run. Then four deposit wallets feed Tornado.
That structure is not improvisation. It is the layering phase of the standard three-stage laundering playbook โ placement, layering, integration โ executed with the minimum competence that actually defeats automation. My read: semi-professional. Someone who has read the same Chainalysis blog posts the investigators have.
Now the forensic counterweight. Tornado's privacy rests on zk-SNARK proofs that decouple deposit addresses from withdrawal addresses. But the anonymity set is only as strong as the correlations you cannot erase: timing gaps between deposit and withdrawal, gas-price fingerprints, relayer selection, and denomination-splitting patterns.

That last one is where this attacker leaked. Four hundred clean, sixty-four messy. The 64 either came through a different pool tier or carries an extra hop's worth of timing metadata. Either way it narrows the search space. The most dangerous thing you can do with a privacy tool is use it correctly-but-imperfectly โ that produces a distinctive fingerprint rather than a blurred one.
And the residual 700 ETH is the loudest signal of all. Three readings, none mutually exclusive: the attacker is washing in tranches to stay under monitoring thresholds; part of the haul is stuck pending a counterparty or a key-holder dispute; or this was a multi-party split and not everyone agrees on pace. A thief who leaves 60% of the take in a wallet he already knows is watched is telling you something about his constraints, not his greed.
Code talks, but stories sell. The code here says "competent layering." The story everyone is pushing says "sophisticated state actor." Those are not the same claim, and the evidence only supports the weaker one.
Here is the angle the market is missing. The reflexive take is "hacker uses mixer, funds gone, crypto is a sewer." That take is lazy, and it is also wrong about where the risk actually sits.
The interesting variable is not the theft. It is the delisting. When Tornado Cash came off the SDN list in March 2025, the privacy-tool narrative was supposed to normalize. Instead, the opposite happened. Without sanctions as a bright line, the compliance burden shifted from "are you touching a sanctioned entity?" to "can you prove where these coins have been?" That is a harder question, and it falls on exchanges, not on the protocol. Tornado's legal rehabilitation did not reduce its fingerprint โ it just moved who has to read it.
The second blind spot: we keep measuring these events by dollar value. 464 ETH is roughly $1.4Mโ$1.9M at current prints โ a rounding error against Ethereum's daily volume. Sub-0.1% price impact. If you are trading this, you are trading nothing. Hype decays; utility endures โ and the durable utility here accrues to the tracing firms, the audit shops, and the exchanges that can prove clean provenance. That is the trade. Not the token.

Watch the 700 ETH, not the headlines. If it splits again into fresh deposit wallets, you are watching a professional finish a job. If it sits, you are watching someone who has lost control of the flow โ a counterparty dispute, a key problem, or fear of a KYC choke point downstream. The real question isn't whether this attacker gets caught. It is whether the next one bothers to divide evenly.