A federal appeals panel has rejected Sam Bankman-Fried’s key arguments on appeal, holding that temporary misappropriation of customer assets satisfies the wire fraud statute and that any later appreciation of those assets does not change the analysis. The ruling also dismisses the claim that FTX’s nature as a margin futures trading platform excused limits on customer access to funds, and it affirms the trial court’s handling of the case under Judge Kaplan. Together, the findings underscore how courts are assessing custodial control, customer consent, and asset movement within centralized crypto platforms.
Technology Overview
At the center of the dispute is how a centralized crypto trading venue manages customer assets and what customers reasonably expect will happen to their deposits. While blockchain networks record transactions on public ledgers, centralized exchanges typically pool user deposits into custodial wallets and maintain an internal ledger to reflect each account’s balances and positions. Customers interact with this internal system to trade, borrow, or post collateral. Because the operator controls private keys to custodial wallets and the database that tracks balances, its technology stack and operational policies directly determine how user assets are safeguarded, segregated, and deployed.
In this case, the appellate panel concluded that the legal concept of misappropriation applies even when an operator asserts that the transfer of funds was temporary or destined for investments expected to grow. The court emphasized that any promise to repay later—or the possibility that assets may increase in value—is irrelevant to whether customer property was taken without proper authorization or under false pretenses. This framing is consequential for crypto infrastructure because it places the focus on the moment assets move and the user’s consent to that movement, not on downstream performance or the operator’s stated intentions.
How It Works
Margin and futures products on centralized platforms rely on an internal risk engine. Users who opt in to margin trading typically post collateral that the system can lock, net, or liquidate under predefined rules. Technically, this means the platform’s matching engine, collateral scheduler, and balance service may flag certain funds as reserved, preventing immediate withdrawal during open positions or until maintenance margin thresholds are met. Users accept these constraints when they choose margin features, and the restrictions apply only to assets they have knowingly pledged.
The appellate ruling drew a sharp line between that kind of voluntary, rules-based collateralization and an operator moving customer assets under false pretenses. The panel said some customers may have chosen margin features and accepted the possibility of temporary limits on access, but others did not. Crucially, no customer consented to funds being transferred to an affiliated entity under misleading circumstances. Translating this into the operational layer of a trading platform: opt-in margin controls are executed by code paths users activate; a transfer that bypasses those expected controls is a different event altogether. One is a feature a user selects; the other is an unauthorized reallocation of custody.
The panel also addressed the argument that eventual gains would offset any interim use of customer assets. In the architecture of a centralized exchange, the balance a user sees is a claim on specific reserves the operator must maintain. Whether a separate investment might appreciate has no bearing on whether the original claim was honored at all times. The court’s analysis mirrors this technical reality: a ledger entry that says funds are available should map to reserves that are actually held for that user, not redirected based on the operator’s assessment of future upside.
Industry Impact
The decision clarifies how legal standards interface with custodial crypto technology. For platforms, the ruling reinforces that customer consent is not a general blanket for asset movement; it must be tied to specific, disclosed features that users actively select. Margin-related deprivation of access is a known, bounded condition governed by transparent rules. By contrast, moving assets to another entity under false pretenses violates the expectations created by custodial balances and the platform’s own representations.
For users, the ruling highlights why the difference between opting into a product feature and trusting a custodial operator’s broader discretion matters. On a technical level, centralized systems can enforce fine-grained permissions: funds pledged as collateral may be locked, while unpledged balances remain withdrawable. The court’s reasoning aligns with that model, distinguishing between voluntary constraints that are integral to how a margin engine manages risk and unauthorized transfers that fall outside any consented workflow.
The panel also supported the trial court’s approach, signaling deference to how the case was managed. This includes the view that intent to repay later does not negate intent to defraud at the time funds are misused. In practical terms for crypto platforms, that increases the importance of rigorous custody controls, clear audit trails within internal ledgers, and governance processes that prevent off-policy transfers. The finding that later asset appreciation is irrelevant further discourages any reliance on after-the-fact portfolio performance to justify how customer funds were handled at the outset.
Future Implications
Looking ahead, the ruling sets a reference point for how Web3 businesses that rely on centralized custody should design and communicate their systems. Platforms offering margin or futures need to ensure that:
- User consent is specific and feature-bound, with technical enforcement that matches what customers were told.
- Internal ledgers and custodial reserves remain aligned, so that an account balance corresponds to immediately available assets unless the user has explicitly opted into constraints.
- Any transfer pathways to affiliates or external entities are governed by transparent, user-facing rules rather than discretionary decisions.
The court’s treatment of the arguments presented on appeal also echoed the reception the defense saw in a hearing last November, when a three-judge panel repeatedly pressed the attorney representing Bankman-Fried. That continuity suggests a stable judicial view on the central issues: temporary misappropriation still triggers the wire fraud framework, and customer consent tied to margin features cannot be stretched to cover undisclosed transfers elsewhere.
For the broader crypto-technology landscape, the message is straightforward. The custody layer is not merely an operational detail; it is the backbone that translates on-chain value into tradable balances for users. Whether an exchange uses sophisticated risk engines, complex wallet orchestration, or scalable internal databases, those systems must faithfully implement what customers were promised. If assets are rerouted in ways customers did not authorize, the presence of trading features, expectations of growth, or later gains do not alter the underlying assessment. The appeals panel’s ruling puts that principle into sharp relief for any organization building or operating infrastructure at the intersection of blockchain and centralized custody.
Finally, the decision’s alignment with the trial court’s actions under Judge Kaplan confirms that the boundaries between agreed-upon margin features and unauthorized asset movements will be enforced as distinct categories. In practice, that means technology, policy, and disclosure must converge: systems should encode what the platform promises, and operations should never assume that future performance can retroactively justify how customer funds were handled. In a sector defined by programmable finance and digital asset custody, the ruling serves as a reminder that code, controls, and consent must match at every step.
