OceanAltOceanAlt

OceanAlt Industry Report · 2026-W43

OceanAlt Agent Economy Industry Watch | Oct 19–25, 2026

This week, the shift from AI-assisted decision-making to autonomous transactions became increasingly evident. As multi-application workflows, agent wallets, and machine payment services continue to advance, competition is expanding beyond payment execution to encompass identity verification, authorization management, and transaction risk controls. Major payment networks, including Visa and Mastercard, are advancing their agentic payment initiatives, while companies such as Tether are increasing investment in the emerging ecosystem. Know Your Agent (KYA) has also gained attention as a key industry topic. Meanwhile, security incidents involving AI agent misuse, prompt injection, and transaction-related risks have reinforced the need for trustworthy decision-making mechanisms. The next generation of payment infrastructure will be defined not only by machines' ability to transact, but also by how effectively authorization, risk, and accountability are managed before transactions take place.

Period:2026-10-19 ~ 2026-10-25(UTC)Published:2026-10-26
Share:XLinkedInFacebookTelegramWhatsAppWeibo🖼 Poster

This week

  • ·21 Caijing reported that multiple apps have restricted calls to Doubao's phone assistant, and financial institutions are tightening access for external agents.
  • ·Tom's Hardware reported that a hacker group used Claude and Grok agents to automate attacks on more than 400 institutions in Latin America.
  • ·Mastercard has integrated the x402 protocol and USDC on the Base chain, positioning itself in agent payment scenarios.
  • ·HSBC launched its first stablecoin, RedCoin, on PayMe, covering more than 3.3 million users.
  • ·Ant International launched the KYA standard, issuing verifiable identity credentials to merchant AI agents.

OceanAlt Agent Payment Pulse

Grouped market-activity events

10events

Method / definition›

Distinct market-activity events (funding / partnership / launch / acquisition) grouped with high confidence by OceanAlt's pipeline this period; deduplicated so multiple articles or languages for one event count once. Each is backed by at least one published, verified in-period item you can click to verify. A verifiable lower bound, not a full-market census.

Period:2026-10-19 ~ 2026-10-25

Last updated:2026-10-25

Coverage & method →

Market activity this period (10 deduplicated, click to verify)

This week's news

Multiple apps restrict calls to Doubao's phone assistant as financial institutions tighten external agent access

2026-10-24Policy & regulationChina

21 Caijing reported that multiple apps have restricted calls to Doubao's phone assistant, and financial institutions are tightening access for external agents.

Why it matters: Access for external agents in app and financial scenarios is running into restrictions from platforms and institutions, reflecting rising compliance and security thresholds for agentic access.

Sources:21 Caijing

Hacker group uses Claude and Grok agents to automate attacks on more than 400 institutions in Latin America

2026-10-23Security incidentLatin America

Tom's Hardware reported that a hacker group used Claude and Grok agents to automate attacks on more than 400 institutions in Latin America.

Why it matters: AI agents being used for automated attacks highlights the risk of agent capabilities being abused in real attack chains and the gaps in security defenses.

Sources:Tom's Hardware

Citi hits pause on AI agent access: the caution of a $2.6 trillion bank and industry divergence

2026-10-23Policy & regulationUnited States

Banking Dive reported that Citi's global head of AI said the bank has blocked all agentic access and is requiring a KYA process, in contrast with Amex, which is proactively setting rules for external agents.

Why it matters: A major bank adopting blocks and KYA prerequisites for agentic access shows that the pace of agent access on the banking side is diverging.

Sources:Banking Dive

SEC lists AI and tokenization as systemic market risks

2026-10-23Policy & regulationUnited States

Decrypt reported that in a risk notice, the SEC listed "superintelligence"-type AI systems alongside asset tokenization as sources of systemic risk that could affect financial stability, but gave no timetable or proposed rules.

Why it matters: Regulators are bringing AI and tokenization into the systemic risk lens, signaling rising regulatory attention, but there are no specific rules yet.

Sources:Decrypt

Money20/20 2026: financial services go "invisible" as stablecoins and agent payments take focus

2026-10-22Opinion & research

A Stripe blog post summarized trends from Money20/20 2026, saying financial services are going "invisible" and that stablecoins and agent payments are in focus.

Why it matters: Trend observations from industry conferences show agent payments and stablecoins becoming a main thread in financial services discussions, but this is an opinion summary rather than implementation validation.

Sources:Stripe blog (primary)

Plaid CEO warns AI agent rise could bring a "terrifying year" for financial security

2026-10-22Opinion & research

Banking Dive reported that the Plaid CEO warned at Money20/20 that the rise of AI agents could bring a "terrifying year" for financial security.

Why it matters: A data infrastructure executive publicly warned about the fraud and security risks brought by agents, reflecting industry concerns about risk controls for agent access.

Sources:Banking Dive

JPMorgan and Goldman Sachs stress security-first AI deployment

2026-10-22Opinion & researchUnited States

Banking Dive reported that CIOs at JPMorgan and Goldman Sachs said the stochastic nature of agentic AI makes security and governance a prerequisite for deployment, and that technology has outpaced governance capacity.

Why it matters: Wall Street bank executives placing security governance as a prerequisite for agentic AI deployment shows institutions' cautious approach to agent implementation.

Sources:Banking Dive

Industry moves

Mastercard integrates x402 and USDC on Base, positioning in agent payments

2026-10-24Partnership

Mastercard has integrated the x402 protocol and USDC on the Base chain, positioning itself in agent payment scenarios.

Why it matters: A card network bringing on-chain stablecoin settlement into agent payment rails means payments initiated by AI agents are entering the view of mainstream card network infrastructure.

Sources:Pluang

HSBC launches first stablecoin RedCoin on PayMe, covering more than 3.3 million users

2026-10-23LaunchedGenerally availableHong Kong, China

HSBC launched its first stablecoin, RedCoin, on PayMe, covering more than 3.3 million users.

Why it matters: A major bank embedding its own stablecoin in an existing consumer wallet is a key step toward regulated stablecoins entering everyday payments; the 3.3 million user figure is disclosed by HSBC, with no independently verified actual usage scale yet.

Sources:CoinDesk (primary)

Ant International launches KYA standard, issuing verifiable identity credentials to merchant AI agents

2026-10-22LaunchedAnnounced

Ant International launched the KYA standard, issuing verifiable identity credentials to merchant AI agents.

Why it matters: Establishing a verifiable identity standard for AI agents is prerequisite infrastructure for agent payments in terms of accountability and compliance.

Sources:Mobile Payment Network

Tether leads Mural's $15 million Series A, betting on AI agent payment infrastructure

2026-10-21Funding

Tether led a $15 million Series A round for Mural, with the funds going to AI agent payment infrastructure.

Why it matters: A stablecoin issuer entering the AI agent payment track through capital reflects its strategic intent to extend into the payment application layer.

Sources:CoinDesk (primary)

Policy & regulation

ERC-8004 draft proposes identity and reputation registry for AI agents

2026-10-24Opinion & researchUnspecified

TOKENPOST reported that the ERC-8004 draft proposes an identity and reputation registry for AI agents.

Why it matters: If the standard advances, identity and reputation records for AI agents could gain a unified on-chain specification, but at the draft stage it has no implementation effect.

Sources:TOKENPOST

SEC lists AI and tokenization as systemic market risks

2026-10-23Policy & regulationProposedUnited States

In a risk notice, the SEC listed "superintelligence"-type AI systems alongside asset tokenization as sources of systemic risk that could affect financial stability, but gave no timetable or proposed rules.

Why it matters: The SEC naming AI and tokenization in a risk notice sends a regulatory attention signal, but there is no rule text or timetable yet, so it does not constitute substantive regulatory action.

Sources:Decrypt

OceanAlt's take

Dense moves in the agent identity layer within a week, but no verified case of cross-platform mutual recognition yet

OceanAlt viewIdentity and authorization credentials for AI agents are becoming the prerequisite infrastructure for agentic payments. This week at least four institutions entered from different paths: Ant International's KYA standard, Sumsub's KYA module, Proof and Circle's x401/x402 end-to-end demo, and ServiceNow and Nvidia's AI Passport. But these solutions belong to different issuers and verification domains, and no public case yet proves that any credential can be mutually recognized across issuers. The identity layer is currently being built in parallel rather than converging.

Evidence: On October 22, Ant International released a KYA identity standard for merchant-side AI agents, claiming credentials can be verified across platforms; on October 20, Sumsub launched a KYA module to control agent access permissions; on October 19, Proof and Circle announced a Money20/20 demo of an end-to-end flow in which x401 presents credentials and x402 completes payment; on October 21, ServiceNow and Nvidia jointly released AI Passport and Arc governance services.

Counter-evidence: If any of the above standards announces a mutual recognition agreement with another issuer or third-party verification integration in the next cycle, the "parallel construction" judgment needs revision; moreover, merchant-side agent identity and consumer-side agent identity may follow different paths, and a single standard may not cover both ends.

Watch: Whether credential mutual recognition announcements appear between two or more independent issuers; whether non-initiating enterprises publicly integrate a KYA solution and disclose integration scale; whether the Money20/20 demo turns into a publicly callable production interface.

Clear divergence among banks on agent access: Citi blocks, Amex sets rules, FDIC flags deposit movement

OceanAlt viewTraditional financial institutions' attitudes toward AI agents accessing accounts are stratifying rather than shifting wholesale. Citi chose to block all agentic access and requires a KYA process to be established first, while Amex pre-sets rules for external agents, and the FDIC chairman lists agents moving deposits as a core current concern—the regulatory focus has shifted from "whether access is possible" to "who authorizes it, how much, and how it is logged."

Evidence: According to an October 23 report, Citi's global head of AI said all agentic access has been blocked and a KYA process is required, contrasting with Amex actively setting rules for external agents; on October 20, FDIC Chairman Travis Hill said at an Axios event in Las Vegas that AI agents moving deposits on behalf of customers is a key issue currently under consideration by regulators and the banking industry; during the same period, JPMorgan refused to open accounts to agents, and multiple financial institutions refused to let external agents operate accounts or initiate transactions on their behalf.

Counter-evidence: Citi's statement is a single executive's public remark and does not disclose whether an internal timetable or exception scenarios exist; if it later announces a phased access plan, "blocking" should be understood as a transitional state rather than a final position. Lightweight consumer-facing scenarios (such as balance inquiries) may not be within the scope of the block.

Watch: Whether Citi publishes specific requirements for the KYA process and a timeline for lifting the block; whether the FDIC issues any formal document or request for comment; whether other large banks follow with blocks or with rules in the next cycle, forming a countable camp distribution.

Card networks move agent authorization upstream to the issuing side, with enterprise payables becoming the first real-money testing ground

OceanAlt viewVisa and Mastercard this week both placed agentic payments within existing card networks rather than building new independent channels: Mastercard opened early access to Agent Pay for Machines and demonstrated a single budget paying multiple service providers, while Visa partnered with BILL and Billtrust to integrate agents into enterprise invoice payables. The direction is to embed authorization, credentials, and settlement into the issuing side, but both remain in pilot or early-access stages with no transaction volume disclosed.

Evidence: On October 20, Mastercard opened early access to Agent Pay for Machines, aimed at autonomous agents purchasing digital services; on October 21, Mastercard separately announced a technical demo showing an agent paying multiple service providers separately under a single budget framework; on October 21, Visa, BILL, and Billtrust launched a pilot for AI-agent B2B invoice payments, pushing authorization, reconciliation, and supplier portal integration into real payables workflows.

Counter-evidence: "Early access" and "pilot" do not equal commercial adoption; if no participant counts, processed transaction counts, or settlement amounts are disclosed in the next cycle, the pace of progress cannot be judged; the payment terms and reconciliation complexity of enterprise payables scenarios may significantly slow the transition from pilot to routine operation.

Watch: Whether the Visa/BILL/Billtrust pilot discloses the number of participating enterprises and invoices processed; whether Mastercard Agent Pay for Machines moves from early access to general availability and publishes pricing; whether integration appears between issuing-side agent authorization and third-party KYA standards.

Agentic-payment security incidents and regulatory risk warnings appear in the same week, with the attack surface extending from the protocol layer to agents themselves

OceanAlt viewRisk signals this week concentrated in two areas: the x402 SDK was found to have a severe vulnerability that could expose agents to malicious interfaces when calling external APIs; according to Straiker research, hacker groups used agents such as Claude and Grok to launch automated attacks on more than 400 banks and payment institutions in Latin America. Meanwhile, the SEC listed "superintelligent" AI systems alongside asset tokenization as sources of systemic market risk, but gave no timetable or proposed rules.

Evidence: On October 23, according to SecNews.gr, the x402 SDK has a severe vulnerability that could expose AI agents to malicious interface risks when calling external APIs; on October 21, The Information reported Straiker research saying CyberCartel used Claude and Grok agents to attack more than 400 financial institutions in Latin America; the SEC, in a risk alert, listed superintelligent AI systems alongside tokenization as sources of systemic risk that could affect financial stability, without proposing rules.

Counter-evidence: The disclosure source for the x402 SDK vulnerability is a single technical media outlet, with no CVE number or official fix announcement yet corroborating the scope of impact; the Straiker research is exclusive research from a single security firm, with attack scale and attribution methodology not public, and the figure of more than 400 institutions should not be directly extrapolated to the industry's overall exposure level.

Watch: Whether the x402 SDK issues a numbered security advisory and fixed version, and the proportion of affected downstream projects disclosed; whether Latin American regulators or the banks involved issue incident confirmations; whether the SEC risk alert moves into a comment period or staff guidance stage.

Period 2026-10-19 ~ 2026-10-25 (UTC). Market data reflects verified metrics you can check at the source; judgments are human-reviewed before publishing. Global industry lens by default.

📬 Get every report in Telegram → subscribe

OceanAlt — Bringing AI agents safely into the real financial world.

TRY IT · FREE, NO SIGNUP

Paste a payee address before you pay and see whether it's on a sanctions list, through a mixer, or tagged for fraud.

This judgement can sit inside your own product

One line of code; it touches neither your CSS nor your JS. The same pre-settlement judgement can appear in your articles, on your wallet's confirmation screen, or as an endpoint your agent calls before it pays.

Get the snippetAbout the Partner Network →

The widget collects no reader identity. Integrating does not mean OceanAlt endorses your product, or any address on your page.