When money becomes programmable, the weaknesses in the system are amplified at machine speed. An AI agent that treats stablecoins as cash equivalents will make perfectly rational decisions based on the information it has access to.
Original text: When Money Moves Itself, Trust Must Be Computable (forbes)
Author: Anusha Nerella, Senior Financial Technology Engineer, focusing on building secure, resilient, and sustainable financial systems for global markets.
Compiled by: Bilibili News
The views expressed in this article represent the author's personal opinion only and should not be used as the basis for investment decisions, nor do they constitute any investment advice. Readers are advised to strictly abide by local laws and regulations.
According to data from CoinDesk Research, the market capitalization of stablecoins has surpassed $300 billion, nearly six times the less than $50 billion at the beginning of 2020.
According to data from Stablecoin Insider (compiled by Market.us News), the annual trading volume of stablecoins reached $33 trillion in 2025, a year-on-year increase of 72%.
As of January 2026, Visa's stablecoin settlement volume had reached an annualized operating level of $4.5 billion.
These metrics are no longer just a topic of discussion within the crypto space. They have become part of the financial infrastructure, which is facing a problem that has yet to be clearly named.
More and more financial transactions are no longer initiated by humans, but completed by software. AI agents are automatically making payments to suppliers, adjusting cash reserves, replenishing accounts, settling invoices, and allocating liquidity between different platforms—all without human intervention.
According to Gartner's forecast, within the next two years, 33% of enterprise software will have "agentic AI" capabilities. Platforms like Pay3 already provide infrastructure designed specifically for AI agents, enabling them to autonomously execute stablecoin payments.
Coinbase has launched x402, a payment standard native to the crypto space that enables smart agents to autonomously complete transactions through smart contracts and automatically settle payments when conditions are met, without human intervention.
Circle, Fireblocks, and Stripe are also pushing for programmable money management, delegating financial processes to AI agents through embedded control mechanisms. The technological infrastructure to support automated finance is in place. However, the infrastructure to support its transparency is lacking.
This gap between the real-time flow of funds and the slow disclosure mechanisms of humans is known as the visibility gap. I believe that in 2026, this is quietly becoming the most critical risk (and opportunity) in the stablecoin space.
Why the board should focus on this now
When money becomes programmable, the weaknesses in the system are amplified at machine speed. An AI agent that treats stablecoins as cash equivalents will make perfectly rational decisions based on the information it has access to.
However, if the latest proof of reserves is a monthly PDF report released two weeks ago, then the agent is actually making decisions based on a snapshot of reality from two weeks ago. It cannot determine whether the reserve structure has changed, whether the redemption mechanism has encountered untested situations, or whether the liquidity of a particular counterparty has changed. It will simply execute.
Standard Chartered estimates that by 2028, US banks could lose up to $500 billion in deposits due to stablecoins, as core banking operations migrate to stablecoin systems. The CEO of Bank of America has also warned that if yield-bearing stablecoin structures are allowed, their adoption could attract even larger inflows of funds.
At this scale, the difference between machine-readable reserve information and periodic PDF disclosures is no longer just a technical detail, but a dividing line for systemic risk.
Transparency is not the same as verifiability.
Progress in information disclosure is commendable. For example, Circle discloses its USDC reserves weekly and receives monthly third-party audits from one of the Big Four accounting firms in accordance with AICPA assurance standards. This is significant progress. However, an audit at any given point in time is like taking a picture of a car in motion.
Regulators are gradually converging on this standard. Europe's MiCA regulations have established uniform transparency and regulatory rules for stablecoins. The Financial Stability Board (FSB) warns that differences in implementation could create opportunities for regulatory arbitrage.
In the United States, the GENIUS Act established a federal regulatory framework in 2025, but it is not expected to be fully implemented until January 2027—during which time the visibility gap will widen further if the industry does not take the initiative.
The trend is very clear: if stablecoins are to begin functioning as currency, they must provide currency-level visibility, and continuously and in real time.
The market will need a three-tiered structure
For technologists designing stablecoin infrastructure or businesses building automated funds management and payment processes, the challenge boils down to three verifiable layers.
The first layer is: the truth of solvency .
In other words, the system can confirm in real time whether the reserves actually exist, whether they conform to the claimed composition, and whether the redemption rules can be executed in the established manner.
This goes beyond regular disclosure. It means that the stored data must be machine-readable and queryable, not just downloadable files.
The emerging proof-of-reserves model in the industry is moving in this direction, but it has not yet become a standard.
Prior to this, any AI agent that viewed stablecoins as "real-time cash equivalents" was essentially operating based on an assumption.
The second layer is: the transaction truth, which can also be called " decision provenance " .
When an automated agent initiates a payment, the system must not only be able to answer what happened, but also explain who (or what) authorized the transaction, which policy it was based on, under what constraints it was executed, and which circuit breakers were used.
Currently, most stablecoin transparency mechanisms are designed with humans as the primary users. However, in an environment dominated by intelligent agents, these records need to be understandable by machines and used for automated compliance checks, rather than just for auditors to review quarterly.
The third layer is: compliance truth .
In other words, it is about proving that a transaction complies with regulatory requirements without revealing sensitive business logic.
Institutions will not run their core financial processes on a system that cannot be audited unless it exposes competitive information.
Whether this layer is done correctly is not just a technical issue, but also a question of whether it can enter the market.
The impending polarization
The World Economic Forum released an analysis at Davos 2026, which quoted the views of Circle's CEO: Stablecoins and AI are converging, moving towards a new internet finance system in which AI can represent users and institutions to conduct transactions globally.
This integration is real and accelerating. But for all of this to function securely, there is a crucial prerequisite—verifiable reality must be embedded in the underlying system from the very beginning. This reality needs to be continuously monitored, like system uptime, and must be understandable to both machines and humans.
The market will diverge. One type of stablecoin offers narrative-based trust: periodic reports, institutional verification, and accessibility for compliance personnel. The other type offers computable trust: facts are verifiable, continuously updated, and machine-verifiable.
Automated intelligent agents can only operate safely within the second type of system. And companies deploying these intelligent agents have no other choice.
Stablecoin issuers that build on verifiable authenticity as infrastructure—rather than simply as a compliance checkpoint to meet regulatory requirements—will become the true engine of automated finance.
The flow of funds depends on speed. What prevents the system from collapsing are verifiable facts.
Disclaimer: As a blockchain information platform, the articles published on this site represent only the personal views of the authors and guests and do not reflect the position of Web3Caff. The information contained in the articles is for reference only and does not constitute any investment advice or offer. Please comply with the relevant laws and regulations of your country or region.
Welcome to the official Web3Caff community : Twitter account | Web3Caff Research Twitter account | WeChat reader group | WeChat official account
