Patent Granted: Before Automating Payments, Automate Authenticity! CycurTrust Secures Patent for 'AI Orchestration for Authenticity Verification' — Essential for Next-Gen AI and On-Chain Finance

CycurTrust Inc. (Headquarters: Shibuya-ku, Tokyo; CEO: Tsuyoshi Suehiro; hereinafter "CycurTrust") announced on July 10, 2026, that it has received a patent grant decision from the Japan Patent Office for Japanese Patent Application No. 2026-088763, confirming the formal establishment of its patent rights. The patent number will be disclosed separately upon official assignment.

This patent covers a technology that enables AI to assess risk on a per-transaction and per-asset basis, and to selectively execute four key processes required for trust (authenticity) verification:

(1) Acquisition of identification information

(2) Collection of evidence

(3) Verification confirmation

(4) Recording of results

The technology precisely controls whether a transaction should be settled or not, based on dynamically selected processes tailored to each case.

The actual verification procedures and their outcomes are recorded on a blockchain as auditable logs. From a societal implementation perspective, CycurTrust refers to this patented technology as "AI Orchestration for Authenticity Verification."

This patent does not replace settlement infrastructure such as stablecoins, tokenized deposits, or other security tokens. Instead, it operates at the stage preceding value transfer, governing critical questions such as: "On what basis is settlement authorized?", "Can settlement be halted when evidence is insufficient?", and "Can the decision-making process be explained after the fact?"

Chapter 1: AI Can Automate Payments—But What Does It Verify Before Paying?

The "Next-Generation AI and On-Chain Finance Vision" proposal by the Digital Society Promotion Headquarters of the Liberal Democratic Party's Policy Research Council (May 19, 2026) envisions a future where economic activities—such as selecting, purchasing, paying, contracting, and obtaining financing—become automated, interconnected, and available 24/7 through the proliferation of AI agents (※1).

One of the future scenarios outlined in the proposal is within manufacturing. The moment a component is delivered, an AI agent verifies quality, quantity, and contract terms against inspection data and automatically settles the payment on-chain using a yen-denominated stablecoin—a world now becoming a reality.

However, there is one critical gap:

"While payments can be automated, the 'verification' process that should precede them remains unautomated!"

Each transaction involves different assets, counterparties, authorities, contract terms, required evidence, sensitive information, and cyber risks. Applying a single, fixed verification process to all cases leads to three major problems:

◆ Insufficient verification

◆ Unnecessary collection of personal data or trade secrets

◆ Continued use of outdated verification paths even when circumstances change

In an era where AI agents move value faster than humans, it is not enough to simply automate payments. We must also mechanically control: "What was verified?", "Why were certain pieces of evidence selected?", "Which verifications were skipped?", "Was the process halted when evidence was lacking?", and "Can the decision-making process be explained afterward?"

In its April 2026 press release, CycurTrust advocated for the necessity of a "Trust Foundation"—where AI agents verify counterparties, authorities, provenance, and condition fulfillment before moving value (※2). This new patent represents the formal protection of a "Control Foundation" that dynamically optimizes this trust foundation on a per-case basis.

Chapter 2: How the Patent Works—AI Automatically Composes the Right Verification Path for Each Case

Under this patented technology, when an authentication request is received, the AI first analyzes the requester, request frequency, asset type, and past authenticity assessment history to determine the risk classification of the case. Based on this, it dynamically composes a "verification path" for the four processes listed above—(1) Identification acquisition → (2) Evidence collection → (3) Verification confirmation → (4) Result recording—determining what to execute, in what order, to what extent, and what to omit for each individual case. The specifics of each process are as follows:

(1) Acquisition of Identification Information

The first process involves acquiring "identification information" to uniquely identify the asset in question. This includes asset identifiers, wallet information, DID (Decentralized Identifiers), VC (Verifiable Credentials), signature data, and asset type. The scope extends beyond physical assets to include non-physical assets such as data and software, as well as hybrid assets like digital twins and RWA (Real World Assets).

The AI selects which identification information to acquire and to what depth, based on the case's risk classification. In low-risk cases, it avoids collecting excessive data, while in high-risk cases, it performs deeper identification to ensure accurate asset verification.

(2) Collection of Evidence

The second process is the acquisition of "evidence" to substantiate authenticity (trust). The AI selectively gathers only the evidence necessary for the case from sources such as blockchain records, manufacturing and supply chain provenance, inspection results, contract terms, and trust anchor (third-party certification) data.

For cases requiring high confidentiality, the system can opt for "zero-knowledge proofs"—a cryptographic technique that verifies qualifications, authorities, or ownership without disclosing the underlying data—instead of collecting wallet or VC information directly. The architecture also supports evidence collection via "QKD (Quantum Key Distribution)" sessions, depending on the level of quantum computing threat. The design prevents both over-collection and under-collection of evidence on a per-case basis.

(3) Verification Confirmation

The third process is "verification"—cross-checking the acquired identification information (1) and evidence (2). The AI executes a sequence of checks generated dynamically, including identifier validation, wallet verification, DID/VC validation, signature verification, and cross-referencing with external records. It also performs consistency checks to confirm that data about physical assets, non-physical assets, and hybrid assets all refer to the same underlying asset or right. Depending on the quantum threat level, the system can employ "PQC (Post-Quantum Cryptography)" for signature verification.

Crucially, the system controls not only what is executed but also what is explicitly excluded. If a valid verification path cannot be constructed by connecting necessary checks, the system does not force an incomplete verification. Instead, it halts execution entirely by marking certain processes as non-executable.

(4) Recording of Results

The fourth process is "recording the results." The system records not only the authenticity determination outcome but also the identifiers of the actual verification procedures used, linking them to the result as an immutable audit log on the blockchain. Any processes skipped in step (3) and the reasons for skipping them are also recorded. When circumstances change, the verification procedure is regenerated, and the system maintains a "reconfiguration history" linking the old and new procedures.

This enables financial institutions, regulators, and auditors to review an immutable, tamper-proof record of what the AI verified, what it did not verify, and why it reached its decision (zero-trust principle). For the audited party, this provides a verifiable trail to fulfill their accountability obligations.

"If verification cannot be completed, do not settle. And the fact that settlement was not performed must also be explainable afterward!"

That is the