AEREDIUM wins patent for AI stablecoin reserve audit

AEREDIUM Holdings received U.S. Patent No. 12,694,407 B1 on July 29 for a four-model artificial intelligence system designed to continuously audit digital-asset and stablecoin reserves in real time. The patent lists Albert Dadon, founder and CEO of AEREDIUM, as the inventor and was issued by the U.S. Patent and Trademark Office.

The system uses four independent AI models that each assess reserve data. At least three of the four models must agree to produce a verdict; a separate compliance engine can override the ensemble. Audit outcomes are reported on a status scale from PASS to HALT. When the system issues a HALT, connected smart contracts can be instructed to reject pending token transactions until reserves are resolved.

Audit findings are recorded to a primary blockchain within minutes and are cross-anchored to a second independent blockchain to create an immutable public record. AEREDIUM said the patent was granted without narrowing of the original claims during examination, which the company says provides broad legal protection for the application of machine-learning systems to autonomous reserve auditing.

Albert Dadon, AEREDIUM’s founder and CEO, said, “Trust in this industry has always been retrospective — you found out whether the reserves were there after it mattered. This patent makes trust a live property of the asset itself. The verification travels with the token.” He added that the architecture is intended to reduce the lag between reserve changes and public reporting.

The company emphasized the system is not a substitute for legally required accounting audits or issuer responsibility for reserves. AEREDIUM stated that issuers must still run any audits regulators require and described continuous verification as complementary to traditional attestations.

AEREDIUM said the patent covers different operational cadences and time windows and may prevent rivals from using off-the-shelf AI tools to perform continuous reserve audits that fall within the granted claims. The firm is preparing international patent filings and engaging with U.S. regulators as it expands the technology following its seventh patent grant.

The architecture is built to handle practical data problems. Custodial bank APIs and other off-chain feeds can be slow or intermittent, so the system uses the freshest available data from each source and treats stale or diverging feeds as risk signals rather than outright failures. That approach keeps the audit cycle running while flagging anomalies for further review.

The technology provides near-real-time public verification of reserve status and can link audit outcomes to automated transaction controls. Developers say continuous monitoring could reduce gaps left by periodic attestations when reserve balances fluctuate between reporting windows.