AEREDIUM secures patent for AI auditing system that replaces quarterly stablecoin checks

AEREDIUM's new patent wants to replace quarterly stablecoin audits with AI that never stops watching

The USPTO has granted AEREDIUM a patent for autonomous reserve auditing using four independent AI models that must reach consensus before issuing a verdict. When reserves breach thresholds, the system can instruct a smart contract to reject pending transactions. Verdicts are recorded on two independent blockchains. It is the company's seventh U.S. patent and targets stablecoins first, with broader applications for any reserve-backed digital asset. The stablecoin industry has long relied on a familiar trust model: accounting firms periodically review reserve balances and publish attestations weeks after the underlying data was collected. For an asset class that moves around the clock, that approach increasingly looks outdated. Australian infrastructure company AEREDIUM believes artificial intelligence can fundamentally change that model. The company announced that the U.S. Patent and Trademark Office has granted U.S. Patent No. 12,694,407 B1, titled "Autonomous Auditing of Digital Asset Reserves Using a Multi-Model Architecture," on July 29, 2026. Invented by AEREDIUM Founder and Chief Executive Officer Albert Dadon, the patent covers a multi-model AI architecture designed to continuously verify digital asset reserves, record audit verdicts on-chain, and automatically enforce predefined reserve integrity rules when necessary. Rather than treating reserve verification as a monthly or quarterly event, AEREDIUM's approach continuously evaluates reserve integrity and records its findings on-chain, creating what the company describes as a live, cryptographically verifiable audit trail. Four AI Models Instead of One At the core of the patented system are four independent AI models, each responsible for analyzing reserves from a different perspective. One specializes in fraud detection, another identifies unusual or previously unseen behavior, a third forecasts reserve health before problems emerge, and a deterministic compliance engine verifies every decision against predefined regulatory rules. Instead of relying on a single model, the system requires consensus between multiple models before reaching a verdict. At least three of the four AI models must agree before a decision is reached, while the compliance engine has absolute veto authority and can override the other models if regulatory conditions are violated. According to AEREDIUM, the goal is to reduce the risk that a single AI failure, hallucination, or manipulation could compromise the integrity of the audit process. From Reports to Real-Time Enforcement Perhaps the most notable aspect of the patent is that the system is designed to do more than generate reports. When reserve conditions deteriorate beyond acceptable thresholds, the platform can automatically instruct a smart contract to reject pending token transactions. Rather than discovering reserve issues after the fact, the system aims to prevent additional activity until the issue is resolved. The system issues one of four verdicts, PASS, CAUTION, ALERT, or HALT, with a HALT triggering the automatic rejection of pending token transactions. Every verdict is permanently recorded to a blockchain during the audit cycle and anchored to a second independent blockchain, creating an immutable public record that cannot be edited or erased, including by the issuer itself. The company says audit cycles are measured in minutes rather than months, allowing users to verify the state of reserves at the time they hold or transact with a digital asset. Why It Matters Reserve transparency has become one of the defining issues in digital assets over the past several years. While many issuers publish attestations from established accounting firms, those reports remain historical snapshots rather than continuous verification. In rapidly moving markets, reserve positions can change long before the next attestation is released. AEREDIUM argues that continuous AI-powered verification could become a new infrastructure layer for tokenized financial assets, particularly as institutions expand their use of stablecoins and other reserve-backed digital assets. "This patent makes trust a live property of the asset itself," said Albert Dadon, Founder and CEO of AEREDIUM. "Instead of asking users to trust periodic reports, the system continuously verifies reserve integrity and permanently records every verdict on-chain." Looking Beyond Stablecoins Although stablecoins are the most immediate application, the underlying technology is designed for any digital asset backed by reserves or collateral. As tokenized deposits, money market funds, treasury products and other real-world assets continue moving onto blockchain infrastructure, demand for continuous verification may grow alongside them. The patent is the seventh U.S. patent in AEREDIUM's intellectual property portfolio and, according to the company, forms a foundational component of its institutional settlement infrastructure, which combines hardware, cryptography and autonomous oversight to secure digital asset transactions. Whether AI-powered auditing ultimately becomes an industry standard remains to be seen, but AEREDIUM is betting that trust in digital assets will increasingly depend on continuous verification rather than periodic disclosure.

AEREDIUM Granted US Patent No. 12,694,407 for Autonomous AI Auditing of Digital Asset Reserves

Austin, Texas, July 30th, 2026, Chainwire Four independent AI models, one verdict, recorded permanently on-chain - replacing the quarterly audit with a continuous one. AEREDIUM Holdings Inc. today announced that the United States Patent and Trademark Office has granted US Patent No. 12,694,407 B1, "Autonomous Auditing of Digital Asset Reserves Using a Multi-Model Architecture," on 29 July 2026. The patent was invented by AEREDIUM founder and Chief Executive Officer Albert Dadon. Today, holders of stablecoins and other reserve-backed digital assets rely on attestations produced monthly or quarterly by accounting firms: a snapshot, already weeks old on publication, of reserves that move by the second. Every major failure in the digital asset industry has lived in that gap between what a reserve was said to hold and what it actually held. No single model can be fooled into a false verdict, because no single model decides. At least three of the four must agree, and the compliance engine can override them all. The system's verdicts run from PASS through CAUTION and ALERT to HALT - and a HALT is not a report but an act: the system automatically instructs the smart contract to reject pending token transactions, stopping an impaired reserve from continuing to trade against its holders. Every verdict is recorded to a blockchain within the audit cycle and anchored to a second, independent chain, creating a permanent public record that no party - including the issuer itself - can edit or erase. Audit cycles are measured in minutes, not months. Any holder of a digital asset can verify the state of the reserves behind it at the moment they hold it. "Trust in this industry has always been retrospective - you found out whether the reserves were there after it mattered," said Mr. Dadon. "This patent makes trust a live property of the asset itself. The verification travels with the token." The patent was examined and granted by the USPTO following substantive examination, and is the seventh United States patent in the AEREDIUM portfolio. It is a foundational component of AEREDIUM's Trust Layer for institutional settlement. About AEREDIUM AEREDIUM Holdings Inc. builds The Trust Layer: blockchain infrastructure for institutional settlement in which correctness is enforced by hardware, cryptography and autonomous oversight rather than promised by intermediaries. AEREDIUM operates from Melbourne, Tel Aviv and Austin, Texas. Contact CEO & Founder Albert Dadon AEREDIUM Holdings Inc. [email protected] Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.