Artificial intelligence is typically framed as a technology story, examined through the lenses of privacy, intellectual property, employment, cybersecurity, or governance. But there is another dimension businesses are only to confront: AI is increasingly becoming a tax issue.

That does not necessarily mean governments will impose a special levy on robots or generative AI tools. The implications are far broader. AI is reshaping how companies create value, deliver services, deploy people and intellectual property, and operate across borders — and it may change workforce requirements and alter traditional business models. At the same time, tax authorities are adopting AI to administration and enforcement.

The scale of that shift is striking. In 2016, only 9% of tax administrations surveyed by the Organisation for Economic Co-operation and Development (OECD) reported using AI. By 2023, that share had risen to 69%, with another 24% in the process of implementing AI solutions — a move from novelty toward the mainstream in under a decade.

The intersection between AI and taxation therefore runs in both directions. Governments must consider how tax systems should adapt to an AI-driven economy, while businesses must operate in an environment where the same technology is already used by tax authorities to assess compliance and detect risks.

AI Is Already a Philippine Tax Issue

The Philippines has already moved on one front, expanding value-added tax (VAT) to digital services — including those supplied by providers without physical presence in the country — through Republic Act No. 12023 and its implementing regulations, in force since June 2025.

The harder question arises when AI is no longer merely a service a company purchases, but a key driver of the value it creates. International tax rules have traditionally relied on concepts such as residence, physical presence, the functions performed, the assets used, and the risks assumed. AI can complicate the application of each of these.

Consider an AI product developed in one country, using intellectual property owned by an entity in another, trained on data from several markets, hosted elsewhere, and sold to customers around the world. Where, then, is the value created — and which government is entitled to tax the resulting profits?

These questions are not easy, but neither are they entirely new. AI amplifies and adds another layer to the puzzles multinational businesses and tax authorities have grappled with for years over profits from intangible assets.

For multinational companies, it may no longer be sufficient to ask which entity legally owns the technology. Tax authorities may also want to know who actually developed and improved it, who controls the relevant risks, where economically significant decisions are made, and which entities perform the functions that ultimately generate income — answers that may be spread across several jurisdictions.

When AI Becomes a Labor-Saving Device

Greater productivity is generally viewed as beneficial. If AI allows an employee to complete in one hour a task that previously took an entire day, the business case appears obvious. The more uncomfortable question is whether those productivity gains will eventually reduce the number of employees required to perform the work.

The question carries particular weight in the Philippines, where labor laws allow employers to terminate employment on grounds that include the installation of labor-saving devices or redundancy, subject to the substantive and procedural requirements imposed by law. The implementing rules provide that labor-saving devices must involve the introduction of machinery, equipment, or other devices in good faith and for a valid purpose, such as saving costs or enhancing operational efficiency. Fair and reasonable criteria must also be used in selecting affected employees.

As AI becomes more capable, Philippine businesses may have to confront a difficult question: could certain AI systems eventually constitute a labor-saving device for purposes of Philippine labor law?

The issue extends beyond employment, because fewer workers may also mean fewer participants contributing to the tax base. Governments derive significant revenues from economic activity associated with human labor. Employees earn salaries and pay income taxes; employment generates mandatory contributions; and workers spend their earnings on goods and services that generate further taxable economic activity.

If a company becomes more productive and profitable while employing fewer people, the state loses some of the revenues associated with human labor — but the economic value does not necessarily disappear. It shifts. Employers may obtain higher margins, technology providers earn revenue, and shareholders may ultimately receive greater returns. The policy question is whether the tax system adequately captures that shift as a growing share of economic value becomes attributable to capital, ownership of intellectual property such as patents and trademarks, and technology, rather than to human labor and the high-value skillsets it requires.

This is one reason discussions of a so-called “robot tax” have emerged internationally — the European Parliament, for one, weighed and ultimately declined to recommend a levy on robots in its 2017 resolution on civil law rules on robotics. Simply taxing AI, however, could discourage investment and productivity-enhancing innovation. The better question is not whether governments should tax the robot, but how they should address the changes AI creates. One approach is to ensure that revenue — even from existing broad-based taxes — is allocated to retrain, upskill, and support regions and sectors immediately affected or undergoing rapid change. The Technical Education and Skills Development Authority (TESDA) is currently offering an AI Certification for the AI-900 exam. There is still a long way to go, but such initiatives can help the terminated labor force adapt and find a place in the changing landscape.

The Taxman Will Use AI Too

Tax authorities will not merely use AI in the future; they are already using it. The OECD reports that AI is being deployed by tax administrations for analytical work, taxpayer services, case selection, and the automation of high-volume, repetitive tasks.

The Bureau of Internal Revenue (BIR) recently introduced automated, risk-based selection for audits that draws data from eFPS and eBIR forms as well as third-party reports such as withholding and import data, and is preparing to roll out the Electronic Invoicing/Receipting system (EIS). How quickly the EIS moves from preparation to rollout is a development Philippine businesses will want to track.

The same capability raises the question of how automated analysis should interact with taxpayers' rights. If an algorithm identifies a company as a high-risk taxpayer, the company may reasonably ask why. It may also need to know what information the system relied upon, whether that information was accurate, and how much weight a revenue officer placed on the system's recommendation. These are not merely technology questions; they are questions of due process. An anomaly is not necessarily evidence of tax evasion. Human judgment, transparency, accountability, and an opportunity for taxpayers to challenge conclusions that materially affect them should therefore remain part of AI-enabled tax administration.

As a precaution, businesses should proactively document legitimate corrections when they occur, so that any automated inquiry can be addressed promptly and supported by contemporaneous evidence. Maintaining records, however, is only side of the equation.

Tax Should Have a Seat at the AI Governance Table

Most companies currently approach AI governance through their technology, cybersecurity, privacy, legal, and compliance teams. Tax should increasingly be part of that discussion. Decisions about where an AI system is developed, which entity owns the relevant intellectual property, how affiliates use the technology, how customers purchase the service, and where resulting revenues are recognized may all have tax consequences.

The technology may be new, but the underlying tax questions are not: who created the value, where was it created, and who gets to tax it? As economic activity shifts toward algorithms, data, and intangible assets, can a tax system designed primarily for physical, labor-driven work keep pace? AI is going to make the answers much harder.

---

The views or opinions expressed in this article are solely those of the author and do not necessarily represent those of Cabrera & Co. The content is for general information purposes only and should not be used as a substitute for specific advice.

Mara Angeli Villegas is a senior legal advisor at Cabrera & Co., a Philippine firm of the PwC network. She can be reached at mara.angeli.villegas@pwc.com.

This article first appeared on BusinessWorld.