Bitget's $388 million theft splits cross-chain swap networks over screening

Shevchenko's own error margin of about 10% puts NEAR's refusals somewhere between $45 million and $55 million, or roughly 12% to 14% of what left Bitget ^[1][2]. THORChain's $6.3 million, as Crypto Briefing reported it, is 1.6% ^[3]. A refused swap leaves the coins in the attacker's wallet, and the account does not say where the refused money went next.

What screening can claim as caught, or rather what it can claim as held, is the money that entered NEAR's swaps before detection finished. Of that roughly $669,000, about three quarters was frozen and a quarter escaped ^[4]. THORChain carried about 38 times the sum that slipped past NEAR ^[5].

Shevchenko argued the screening cost the business nothing it wanted. He noted that NEAR routinely processes more than $100 million a day and called the refusals a negligible fraction ^[7]. At that floor the refusals equal at most half of one day's volume, or about 7% of a week's ^[6]. The refused volume was attacker flow in any case. A screening rail pays in two other places: for the outside AML data it checks quotes against ^[6], and in the legitimate users it may stop. Vini Barbosa, a technical writer building at Ramp Labs, wrote on X that permissionless has to mean neutral, and that a rail willing to restrict suspected unlawful users will also restrict people moving money under repressive governments ^[9].

NEAR cofounder Illia Polosukhin offered a narrower definition, in which permissionless means nobody needs permission to own assets, transfer them or deploy contracts ^[10]. Shevchenko said the service will remain "permissionless infrastructure, but with boundaries" ^[8]. THORChain defended itself on X. "A halt is not a selective freeze of specific funds or an individual swap," the protocol wrote ^[11]. Bitget CEO Gracy Chen, whose request THORChain declined, wrote that decentralization "is a design principle, not a shield for facilitating known stolen funds" ^[13].

The split can settle in a few ways. If the refused $45 million to $55 million later crossed THORChain or another neutral rail, screening moved the loss between venues without shrinking it ^[1]. Institutional money following its stated preference for screened rails would instead mean NEAR's data bill buys flow that THORChain gives up ^[16]. And if Barbosa is right, the neutral rail keeps the users a screen turns away, and THORChain's policy has customers of its own ^[9].

I think the product difference is real and the recovery difference, so far, is small. That view is wrong if the refused money is later frozen or recovered downstream. NEAR's refusals would then have bought the time for a freeze elsewhere, and they would be worth far more than the roughly $503,000 SHIELD held mid-swap ^[4].