On September 11th, at the 2026 Inclusion · Bund Conference AI Payment Forum, Ant Group released the APASS Trust Infrastructure for Agent based Business Activities.

Based on the KYA (Know Your Agent) concept, APASS establishes two trust chains for identity and behavior through intelligent agent trusted identity products and intelligent agent risk control services, focusing on "who the agent is, who it represents, what it is allowed to do, and whether its behavior is trustworthy". It provides identity registration, continuous verification, intent security, and trusted authentication capabilities, providing a trust foundation for agents to call services, handle business, and participate in commercial activities such as payments.

As agents move from "tools" to "actors", a new set of business trust infrastructure is becoming an important infrastructure for the AI new economy.

Agents are becoming new participants in the business world

In the past, AI helped people generate content, obtain information, and assist in decision-making more; Now, agents are further performing tasks on behalf of humans.

It can represent users in searching for products, booking tickets, shopping, and even completing payments, as well as representing enterprises in procurement, marketing, customer service, and business collaboration. As more and more business activities are carried out autonomously by agents, the participants in business activities are changing - transactions that were initiated and executed by people in the past are now beginning to show a process where agents act on behalf of people.

Gartner predicts that by 2028, 90% of B2B procurement will be mediated by AI agents, driving over $15 trillion in B2B spending through AI agent trading platforms.

This also brings up a question that did not exist in the past: when AI starts doing business for people, who should the business world trust?

The trust system of the traditional Internet is mainly built around the "person account platform", and the transaction subject is confirmed through real name authentication, account system and risk control mechanism. But Agent is neither a traditional person nor just a software that passively executes instructions. It can perceive, make decisions, call tools, and act autonomously.

Therefore, after entering the real business world, agents must not only address whether they can do it, but also whether they can do it with confidence.

Chen Shupeng, AI Payment Security Manager at Ant Group

Agents are becoming the 'new population' of the business world. When it starts calling services and participating in transactions, the industry needs to answer both 'who is trading' and 'why is trading'. ”Chen Shupeng, the head of AI payment security at Ant Group, stated that APASS helps participants confirm who the collaborating agent is, who they represent, what authorization they have obtained, whether they are currently trustworthy, and provides traceability evidence in case of any issues. ”

From 'Who are you' to 'Who do you represent and what can you do?'

One change brought about by agents is that traditional one-time identity authentication is no longer sufficient.

An agent may be created by an individual but run on a third-party platform; It may represent an individual or a company; It may only be authorized once, or it may continue to call the service. At the same time, its operating environment, task objectives, and actual behavior may also change at any time.

Therefore, in the era of agents, what is needed is not just confirming an ID, but a continuous trust mechanism that runs through the entire process of action.

APASS integrates identity confirmation and behavioral risk management throughout the entire process of Agent participation in commercial activities, focusing on four core competencies: trusted identity, continuous verification, intent security, and trusted authentication.

At the identity level, APASS uses a three-layer model of static identity, runtime identity, and responsible subject identity, combined with a six dimensional continuous verification mechanism, to identify agents, confirm their operational status, and associate them with the responsible subjects behind them.

At the behavioral level, APASS verifies identity and authorization before transactions, identifies abnormal behavior and intention deviations during transactions, implements dynamic risk control, and retains key evidence after transactions to provide a basis for auditing and accountability tracing.

In other words, APASS hopes to further extend the traditional "ID verification" to continuous confirmation of Agent identity, authorization, intent, and behavior.

Behind this is a bigger industry change: computing power enables AI to run, models make AI smarter, and trust mechanisms determine whether AI can truly enter the real business world.

More than 1 million agents have been connected, and over 50 institutions have jointly built an ecosystem

According to on-site disclosure, APASS has currently connected over 1 million agents, of which 110000 have been activated, covering 442 intelligent agent service providers and supporting different types of agents such as independent assistant, productivity, hardware terminal, and merchant service.

Developers can complete access in three steps, embedding trusted capabilities into service invocation, business processing, and payment processes. APASS uniformly adapts to different entities and technology paths, reducing developers' repeated investment in identity access and security capability integration.

At the same time, including Alibaba, Ideal Auto, Ant Group, Damai, Lenovo OPPO、 More than 50 institutions, including Qianwen, Jieyue, Xiaomi, Ctrip, Bank Card Testing Center, Zhipu, Zhima Credit, China Academy of Information and Communications Technology, and China Mobile Jinko, have joined the IIFAA Intelligent Agent Trusted Ecological Co construction Plan to jointly promote the development and implementation of relevant alliance standards.

In terms of international protocol integration, Alipay+AMP, Mastercard verifiable intent, and Visa intelligent agent trusted protocol are used to build a KYA interconnection framework, providing support for agent identity verification and trust collaboration in cross-border business scenarios.

From the transaction trust in the digital economy served by Alipay, to APASS supporting agents to participate in business activities, ants are extending their long-term accumulated identity, security, risk control and payment capabilities to the agent scene.

In the past, Alipay solved how to establish transaction trust between people and between people and businesses; In the future, APASS explores how to establish action trust between people and agents, agents and merchants, and even agents and agents.

When AI starts to move from "thinking for people" to "acting for people", enabling AI to be recognized, authorized, trusted, and take responsibility when problems arise, it may be the layer of infrastructure that agents must fill in as they move from technology to business and from AI to the AI new economy.