Know Your Agent: The New Visibility Challenge for Tax Authorities
Imagine an entrepreneur with a laptop and a few AI tools. Within months, they are operating several businesses at once: a handful of online stores, a content and translation service for overseas clients, a subscription app, and a portfolio of short-term rental listings. AI agents create listings, deliver much of the work, respond to customers in several languages and adjust prices overnight. To customers, these may appear to be small, unrelated businesses operating under separate brands. To a tax authority, the links between them may be completely invisible.
Nothing about this arrangement is necessarily improper. The operator may be reporting all of their income. But the example illustrates an emerging challenge for tax authorities: AI agents make it possible for operators to conduct more economic activity, of more kinds, faster and across more platforms and borders, while leaving fewer of the traditional records and indicators that enforcement has traditionally relied on.
Banks have addressed a similar challenge with a simple principle: know your customer. As more economic activity runs through AI agents, tax authorities may need a complementary approach. At IVIX, we call it "Know Your Agent:" understanding who is behind AI-enabled economic activity, and whether that activity is already known to the authority.
Why AI-Enabled Economic Activity Is Harder to See
AI agents are software tools that can perform business tasks with limited human involvement. They primarily help legitimate businesses operate more efficiently, and their use is not a risk indicator in itself. Nor is fragmented economic activity a new phenomenon. What AI agents can do, however, is amplify its scale, speed and complexity by enabling a single operator run several business models at once. For tax authorities, that changes the external signals, or how economic activity looks from the outside, in three important ways:
• Scale without staff. A very small team can operate a business that previously required many employees, making headcount, payroll and physical premises less reliable indicators of its economic activity.
• Speed and fragmentation. Stores, services, apps and other ventures can be launched in days or even hours, with activity spread across multiple platforms, business models and countries. Each venture may appear insignificant when viewed in isolation.
• Distance from the people behind the activity. An operator can be based anywhere and work through several entities, platforms and payment providers, blurring the link to an identifiable taxpayer.
These changes create practical challenges for tax enforcement. Hidden economic activity is harder to identify and connect for overall significance when split across many small businesses and sectors. Case selection also becomes more difficult when no single record captures the full picture of an operation, potentially directing attention toward the most visible businesses, while larger, interconnected operations stay out of view.
A Hypothetical Case: Four Ventures, One Operator?
Consider our entrepreneur once more, from the perspective of a tax authority. The online stores trade through a registered company, while the content service appears to belong to a sole trader. The app is published under a developer account in another name, and the rental listings appear under a host profile with no visible registration. Each activity sits in a different sector and is too small to attract attention on its own.
From the outside, however, a different picture may emerge. The content service's website and the stores' contact pages display the same business address, while the app's support page and the rental listings share a phone number. Several of the websites were created within days of one another, and a public professional profile identifies its owner as the founder of "AI-run ventures," naming two of the brands.
These clues do not establish common ownership or unreported income. Businesses may legitimately share addresses or service providers, and entrepreneurs often operate multiple ventures. Taken together, however, these signals raise a worthwhile investigative question: are these independent businesses or parts of a connected operation, and does the authority have visibility into the full scope of activity? The challenge is not a lack of information. It is that the relevant information is scattered across sources, sectors and platforms that no single record brings together.
What Remains Visible
Economic activity aimed at customers, clients or users must be visible to them, and that public visibility generally leaves traces, including:
• Storefronts, marketplace and rental listings, service profiles, app pages, and advertisements, along with public reviews or ratings that may indicate activity levels
• Shared details across seemingly separate ventures, including websites, domain registration information, contact details, addresses, branding or support channels
• Public registry records for the entities behind the activity, in one or more jurisdictions
• Public professional profiles and company pages identifying people who operate or promote the ventures
These are public indicators, not transaction records, and no single signal is proof. The value lies in combining and cross-checking them to form a reasoned view of whether activities may be connected, how significant they maybe, and whether a closer look is warranted.
Know Your Agent: A Framework for Tax Authorities
For tax authorities, “Know Your Agent” means answering three questions about AI-enabled economic activity:
• Who operates it? The individuals and entities that appear to control or conduct the activity.
• Who benefits from it? Where the proceeds appear to flow, and to whom.
• Is it known to the authority? Whether the activity is already on the authority's radar, and whether there are indications of a gap between what is visible and what was reported.
Internal tax records can answer some of these questions. Open-source information can help with the rest.
From Open-Source Signals to Actionable Intelligence
This is where open-source intelligence (OSINT) becomes a practical tool for tax authorities. IVIX works with tax authorities to discover economic activity using OSINT, using an approach that addresses the four parts of this challenge:
• Discovery. Identifying economic activity, across sectors, platforms and business models, that may not appear in the authority's own records.
• Scale. Analyzing public indicators of how significant that activity may be.
• Connection. Cross-referencing signals that may link seemingly unrelated businesses, accounts and entities to common operators, and to identifiable individuals or companies.
• Prioritization. Comparing these findings with the authority's own information, to flag potential gaps and surface cases that merit closer examination.
For tax authorities, the practical value is greater visibility and sharper focus: seeing activity that may otherwise stay hidden, knowing which activities may be connected and which cases merit a closer look, and where limited resources can make the greatest difference. These findings are starting points for investigation, not proof of ownership or taxable income; interpretation and enforcement decisions remain with the authority's experts.
Why This Will Matter More
AI agents are becoming more accessible, easier and cheaper to deploy, making broader adoption increasingly likely. Most of this activity will be legitimate. But as more economic activity becomes automated, the gap between what happens online and what appears in official records may widen.
For tax authorities, this is an emerging strategic question, not a distant one. It points to a shift from examining businesses and records in isolation to understanding connected patterns of economic activity across platforms, sectors and entities. Administrations that build that view now will be better prepared and better able to leave compliant businesses undisturbed.
Visibility Comes First
The rules on taxing AI-driven activity will continue to evolve. Whatever shape they take, tax authorities will still need the same starting point to be effective: a clear view of the activity, the people and entities behind it, and the indicators that call for a closer look. AI agents may change how businesses operate, but not the fundamental questions tax authorities must answer. AI agents may do the work, but there is always someone behind them.
Ready to See More of Your Digital Economy?
Discover how IVIX helps tax authorities gain visibility into potentially hidden economic activity across sectors and platforms, connect fragmented signals and prioritize cases for further review.
