Taxing AI: Policy Proposals and IRS Enforcement Reshape the Landscape

The Push to Tax Artificial Intelligence

A growing chorus of politicians and policymakers is advocating for new taxes on artificial intelligence, arguing that the technology's economic benefits must be shared broadly. Proposals range from token-based levies to universal basic capital, reflecting a bipartisan interest in redistributing AI-generated wealth. The urgency is driven by AI's wide-reaching impact on American workers and the fact that many models were trained on human labor.

Senator Elizabeth Warren has been a vocal proponent, stating, "If we overhaul our tax code and tax AI, we can use that money to build a country that works for everyone." Her vision includes funding healthcare, guaranteed jobs, and education. Representative Greg Casar of Texas has proposed a tax on tokens—the units of data processed by AI models—while Michigan State Senator Mallory McMorrow has called for a similar tax on commercial AI uses to fund apprenticeship programs. Senator Ron Wyden is exploring a tax on tech companies to create a wage-security program for displaced workers.

Even some in the tech sector support the idea. Anthropic CEO Dario Amodei has argued that the "extreme levels of inequality" predicted from AI "justify a more robust tax policy on basic moral grounds." This bipartisan interest, including from figures like Bernie Sanders and Donald Trump, signals that the debate is moving from fringe to mainstream.

Universal Basic Capital: A New Redistribution Model

Beyond direct taxation, a concept called "universal basic capital" is gaining traction. This policy would give ordinary people an ownership stake in firms that profit from AI, reframing the debate from income support to capital distribution. Advocates across the political spectrum, including Bernie Sanders, Gavin Newsom, Steve Bannon, Sam Altman, and Donald Trump, have voiced support for variants of the idea.

Implementing such a model requires robust methods to measure AI-attributable revenue and clear legal instruments for fractional claims. Design choices—dividends, equity, or tokenization—will determine whether the policy mitigates inequality or creates new concentration and governance risks. For practitioners, the operational details matter: distributing capital claims tied to AI output requires reliable measurement of model-attributable revenue, new accounting conventions, and mechanisms for fractionalized ownership.

The IRS Codifies AI in Audit Selection

On February 10, 2026, the IRS codified its AI enforcement practices into formal policy with IRM 10.24.1. This new Internal Revenue Manual section establishes what AI is authorized to do in the examination process, how mandatory human review is integrated into AI-assisted case selection, and what documentation requirements apply to AI-generated referrals. The significance is clear: AI is no longer experimental inside the IRS—it is operational and codified.

The IRS now operates two separate AI models to prioritize large partnership returns for examination. For individual returns, AI models select a stratified sample and identify those statistically most likely to contain errors or underreported income. Early pilot results show the new models outperform prior selection methods by a meaningful margin, replacing the older Discriminant Information Function (DIF) model that produced high "no-change" audit rates.

New Liabilities for Tax Professionals

As the IRS embraces AI, tax professionals face new risks. AI systems do not actually understand tax law—they predict language patterns. This predictive capability can produce helpful summaries but also fabricated court cases, oversimplified interpretations, or entirely incorrect tax positions presented with absolute confidence. Without fundamental knowledge, clients can be led astray by these convincing outputs.

Tax professionals are already seeing clients bring in AI-guided strategies. Firms are advised to update client intake questionnaires to ask about new entities or strategies implemented without professional consultation. If a client insists on an AI-supported position, firms should document warnings, potential penalties under Section 6662, and the client's decision to proceed against advice. Clear internal policies on AI use for tax research are essential, with mandatory verification of every citation against primary authority.

IRS AI Enforcement: What Practitioners Need to Know

The IRS's new AI enforcement policy, IRM 10.24.1, removes any ambiguity about whether AI is experimental or operational. It establishes three key things: what AI is authorized to do in the examination process, how mandatory human review is integrated into AI-assisted case selection, and what documentation requirements apply to AI-generated referrals. The human review is structured around AI outputs, not the other way around.

According to a GAO analysis, the IRS now operates two separate AI models to prioritize large partnership returns. For individual returns, AI models select a stratified sample and identify those statistically most likely to contain errors or underreported income. Early pilot results show the new models outperform prior selection methods by a meaningful margin, replacing the older DIF model that produced high "no-change" audit rates.

Professional Liability in the Age of AI

Tax professionals are already seeing clients bring in AI-guided strategies. The danger is that AI systems do not understand tax law—they predict language patterns. This can lead to fabricated court cases, oversimplified interpretations, or entirely incorrect tax positions presented with absolute confidence. Without fundamental knowledge, clients can be led astray.

Firms are advised to update client intake questionnaires to ask: "Have you established any new entities, changed your residency or implemented any tax strategies in the last 12 months without consulting a licensed professional?" If a client insists on an AI-supported position, firms should document warnings, potential penalties under Section 6662, and the client's decision to proceed against professional advice. Clear internal policies on AI use for tax research are essential, with mandatory verification of every citation against primary authority.

Navigating the New Landscape

For firms thinking ahead, the most defensible path is designing around the disclosure question. Running AI on hardware the firm controls means taxpayer information never leaves, eliminating the need for disclosure under Section 7216. The assumption that sending data to a large outside model is the only way to use AI is already becoming outdated.

The firms best positioned to thrive are not necessarily the largest or most technologically sophisticated. They are the ones that understand what the IRS's AI is actually doing, translate that understanding into proactive client guidance, and use this shift as the foundation for deeper, year-round advisory relationships that no algorithm can replicate. The conversation the profession isn't having is about whether existing frameworks for due diligence, competence, and confidentiality are adequate for the AI era.