AEREDIUM Holdings recently obtained a US patent for an artificial intelligence (AI) system designed for the continuous real-time auditing of digital asset reserves, aiming to replace traditional accounting attestations with automated oversight. On July 29, the United States Patent and Trademark Office issued Patent No. 12,694,407 B1. The system was invented by Albert Dadoun, the founder and CEO of AEREDIUM.
Issuers of stablecoins and other reserve-backed digital assets currently rely on monthly or quarterly attestations prepared by audit firms. The publication of such reports often takes weeks, while reserve balances continuously fluctuate, leading to delays that have played a significant role in major digital asset market disruptions.
According to a press statement, AEREDIUM's patented system replaces periodic reports with continuous monitoring using four independent artificial intelligence models. Within this architecture, no single model can determine the outcome on its own. A consensus of at least three out of the four models is required to reach a verdict, with a compliance control mechanism having the authority to reassess their decision.
The system issues status levels ranging from 'PASS' to 'HALT'. Upon issuing a 'HALT' status, the system automatically instructs connected smart contracts to reject pending token transactions, preventing continued trading when reserves are non-compliant. Audit results are recorded on the primary blockchain network within minutes and are cross-referenced to a second independent blockchain network, creating an immutable public ledger.
"Trust in this industry has always been retrospective — you found out whether reserves existed only after it already mattered," Dadoun said. "This patent makes trust an intrinsic property of the asset itself. Verification accompanies the token."
Dadoun noted that the patent was granted without any narrowing of the original claims during USPTO examination, providing the company with broad legal protection across various operational cycles and time intervals.
"The machine learning components in an abstract sense belong to everyone, but when applied to this specific use case — autonomous auditing of digital asset reserves — they are protected," Dadoun said. "A competing issuer or audit firm applying these models to stablecoin reserves would be operating within the issued claims, regardless of where those components were sourced."
He added that AEREDIUM is actively preparing to file international patent applications.
Addressing potential operational hurdles, such as delays in custodian bank API operations, Dadoun explained that the architecture is designed with slower off-chain data flows in mind. The cycle uses the latest available data from each source, treating stale or divergent data streams as risk signals rather than system failures.
Dadoun also clarified that the AI ensemble is not intended to replace legally mandated financial audits or to reduce issuer liability.
"The legal liability for stablecoin reserves lies with the issuer, and whatever audit the regulator prescribes, the issuer is obligated to perform it," Dadoun told Bitcoin.com News, characterizing continuous verification and traditional audits as complementary processes.
However, Dadoun indicated that regulatory expectations may evolve with the adoption of real-time verification tools, noting that AEREDIUM is currently in discussions with US regulators.
"We expect regulators to move from simply acknowledging continuous verification to mandating it," Dadoun said. "In the US, the reserve requirements for payment stablecoins are enshrined in law, and continuous verification against those legally established standards is precisely what this architecture was built to do."
Header/Primary Image Source: SBS Hebrew/sbs.com
