The Coldcard exploit has produced its first visible act of restitution, and the transaction carries its instructions in plain sight. Whitehat operators consolidated 52.37 BTC from tracked exploit clusters into a fresh address in Bitcoin block 967,948, stamped with an OP_RETURN message pointing to "claim:cryptorecoverytrust dot com," routing the funds into a Wyoming statutory trust built to return rescued Bitcoin to verified owners. Galaxy Digital's head of research Alex Thorn identified the coins as coming from the Wave 2 cluster, footprints labeled AA, AU and AX, and calculated the transfer at 2.8% of the exploit funds his team tracks. That figure is a research tally, not an official loss number, and it is small against the overall incident. The mechanism matters more than the amount: it is a working template for how whitehat recoveries can be made auditable, claimable and legally segregated.
Why the Coldcard recovery trust exists
The trust structure is the story's most interesting design. The funds went to the Recovered Digital Asset Statutory Trust of Wyoming, with Agentic Trace LLC as trustee and Steptoe's national security practice advising, because some recoveries may involve sanctions screening, law-enforcement coordination or competing ownership claims. The Digital Asset Recovery Trust, or DART, disclosed in August that it and independent whitehats had secured just over 50 BTC from vulnerable addresses before malicious actors could reach them, and its process runs blockchain analysis, proof-of-ownership checks and sanctions screening before any return, with disputed funds following separate legal procedures. The whitehats requested no bounty, per DART. That is the exact opposite of the standoff Immunefi's CEO described in the Liquid Network case: take nothing, disclose privately, sweep vulnerable funds ahead of thieves, and put the money somewhere neutral while the lawyers sort out ownership.
The origin of all this remains the worst hardware wallet failure of the year. The July 30 exploit began with a firmware integration defect that routed seed generation to MicroPython's Yasmarang pseudorandom generator instead of the hardware random number generator, making affected seeds searchable offline, with public research tracing the weakness to 2021-era firmware changes and estimating roughly 40 bits of effective entropy on older Mk3 devices and about 72 bits on Mk4, Mk5 and Q models. First-wave losses reached roughly 594 BTC from around 500 wallets in about 25 minutes, and expanded tracking has estimated 1,816 BTC moved from more than 5,200 addresses across four attack waves, though totals vary by scope and no single official figure exists. The exploit required no remote takeover: attackers simply regenerated weak keys and swept the funds, the same seed-integrity failure mode that makes legitimate wallet makers obsess over backup and generation design.
What firmware can and cannot fix
Coinkite's line has stayed consistent and blunt: patched firmware fixes future seed generation, but it cannot repair a seed already generated under the vulnerable path. A wallet created on affected firmware remains exposed even fully updated, and users holding such seeds must generate a new one and migrate funds, with the company's one exception being at least 50 fair, independently recorded six-sided dice rolls adding 128 bits of entropy under its documented workflow. The current recommended releases are Mk4/Mk5 version 5.6.2 and Q version 1.5.2Q, both issued Sept. 3, with Edge users directed to 6.6.1X and 6.6.1QX. For anyone who lost funds, the recovery path now runs through the trust: claimants can search for recovery information, submit and track a claim, and supply supporting evidence through the trust's website, with DART noting other vulnerable assets and leads remain under review, meaning more recovered Bitcoin could enter the process.
What to watch
Three markers will tell you whether the 52.37 BTC is a gesture or a system. First, whether additional consolidations follow, since DART's August tally covered just over 50 BTC and the trust model only matters at scale. Second, whether claims processing actually returns coins to verified victims quickly, because a trust that takes years to distribute is cold comfort, and the competing-claims and sanctions carve-outs can swallow small balances whole. Third, whether the pattern spreads, since the industry has lacked a standard for whitehat recoveries precisely because the alternatives, keep the funds, demand bounties, or hand everything to investigators, each broke something. The Coldcard trust is the first big test of the fourth path: sweep early, hold neutrally, verify honestly, return. Two-point-eight percent is a small number. If it becomes the default, it is the most important two-point-eight percent in the year's security record.
