Senator Mark Kelly (D-AZ) has introduced the "Make AI Work for Americans Act," a legislative proposal driven by growing concerns regarding the profound and potentially rapid transformations that artificial intelligence (AI) adoption could inflict upon the labor market. The bill aims to establish federal funding for a suite of labor insurance and worker training programs, designed to cushion the economic impact on American workers as AI technologies become more pervasive. To finance these substantial new expenditures, the Act controversially proposes three distinct new federal taxes: a digital advertising tax, a bit tax, and an excess profits tax, all slated to take effect on January 1, 2027.
The Evolving Landscape: AI’s Impact on the Labor Market
The discussion surrounding AI’s true impact on employment remains a subject of intense debate among economists and policymakers. While some analyses suggest a future of widespread job displacement, others foresee the creation of entirely new roles and a significant boost to productivity across various sectors. The Stanford Digital Economy Lab and Yale Budget Lab are among the institutions actively tracking and debating the nuances of AI’s integration into the workforce. For instance, projections from leading consulting firms like McKinsey indicate that AI may necessitate a substantial overhaul of the workforce, potentially requiring as many as 11 million workers to transition into new job roles or acquire entirely new skill sets. This looming uncertainty, characterized by both promise and peril for the American worker, underpins the perceived urgency behind Senator Kelly’s legislative efforts to prepare the workforce for an AI-driven economy. The proposed federal programs, including enhanced unemployment benefits, vocational retraining initiatives, and educational pathways, seek to mitigate potential economic dislocation and ensure that American workers can adapt effectively to evolving industry demands and skill requirements.
Funding the Future: An Overview of Senator Kelly’s Proposed AI Taxes
The financial architecture of the "Make AI Work for Americans Act" relies heavily on the aforementioned trio of novel taxation mechanisms, each designed to capture revenue from different facets of the digital and AI-driven economy. These proposals represent a significant shift in federal tax policy, targeting specific sectors believed to be benefiting most directly from AI’s advancement and the broader digital transformation.
- Digital Advertising Tax: This tax targets the revenue generated from online advertisements. It is structured as a 5 percent levy on the gross revenues of digital advertising services for companies exceeding $100 million in annual gross receipts from such services.
- Bit Tax: Conceived as a levy on data usage, this tax would apply at a rate of $0.11 per gigabyte (GB) of data transmitted or stored. The proposal includes an exemption for the first 10 GB of usage per user per month, aiming to shield individual consumers from the tax.
- Excess Profits Tax: This proposal aims to tax companies that achieve what are deemed "excess" profits. Specifically, it applies a 25 percent tax rate on profits that exceed 40 percent of a company’s gross receipts.
Initial revenue projections, estimated without accounting for behavioral changes or broader economic growth impacts, suggest varying degrees of fiscal impact in the first year. The digital ad tax is projected to generate approximately $15.8 billion, while the excess profits tax is estimated at $6.5 billion. In stark contrast, the bit tax is anticipated to raise a modest $23.2 million in its inaugural year. Over a ten-year horizon, these three taxes could collectively accrue an estimated $380 billion, though analysts caution that this figure represents an optimistic ceiling, as it does not incorporate potential avoidance strategies, macroeconomic feedback effects, or shifts in economic behavior. These estimates also account for an expected offset against existing income and payroll taxes, acknowledging that new taxes can reduce the base for other federal revenue streams.
Fiscal Responsibility or Future Deficits? A Critical Examination
Senator Kelly’s initiative to identify funding sources for new spending is commendable, especially given the United States’ escalating federal debt burden and the near-record levels of interest payments on that debt. In an era where fiscal prudence is increasingly critical, policymakers are rightly encouraged to propose corresponding revenue streams for new expenditures. However, a significant criticism leveled against the "Make AI Work for Americans Act" by various tax policy analysts is that despite its stated intent, the proposal may inadvertently contribute to increased federal deficits. This concern stems from the bill’s approach to revenue calculation: it accounts for the gross revenues generated by the new taxes but potentially overlooks the expected declines in other existing tax receipts, such as corporate income and payroll taxes, which could result from these new levies. For instance, new business taxes can reduce a company’s taxable income under existing corporate tax rules, thereby reducing federal revenue from those sources. This "net effect" consideration, encompassing both direct revenue gains and indirect revenue losses, is crucial for an accurate assessment of the bill’s true fiscal impact. Without fully accounting for these feedback loops, the promise of self-funded programs could prove illusory.
Historical Precedents and Their Lessons
While Senator Kelly’s specific combination of these three taxes may be novel at the federal level, each concept has distinct historical antecedents at federal, state, or international jurisdictions, offering valuable insights into their potential efficacy and pitfalls. The past serves as a crucial guide for understanding the potential real-world implications of these modern proposals.
The Digital Advertising Tax: A State-Level Battleground and International Precedent
The concept of a digital advertising tax has gained traction most notably at the state level within the United States, although its implementation has been met with considerable legal challenges. Maryland, for instance, became the first state to enact such a tax in 2021, with Illinois and Utah following suit with similar measures targeting online advertising revenue in recent years. However, Maryland’s digital ad tax faced a significant setback when the state’s Tax Court struck it down in August, ruling it unconstitutional. This highly watched case highlighted complex constitutional questions surrounding interstate commerce and the internet Tax Freedom Act, prompting an appeal by the state. This ongoing legal battle underscores the significant hurdles in implementing such taxes at a sub-federal level.
Internationally, numerous countries have implemented or proposed Digital Services Taxes (DSTs), which often share characteristics with digital ad taxes. These DSTs, typically levied on the revenues of large tech companies from digital services within their borders, have frequently sparked retaliatory threats from the United States, which views these taxes as unfairly targeting American tech giants and conflicting with global tax norms. Critics argue that digital ad taxes, particularly those levied on gross receipts, function akin to cascading taxes. They can lead to "tax pyramiding" by taxing transactions at multiple stages of the advertising supply chain (e.g., advertiser to platform, platform to publisher), ultimately increasing costs for businesses and potentially stifling innovation in the digital economy.
The Bit Tax: A Blast from the Past with Modern Implications
The idea of a "bit tax"—a levy on digital data—is not new; it has a history dating back to the 1990s, an era when the internet was nascent. Proponents of the 1990s bit tax envisioned a modest charge per unit of data transmitted, a concept that seemed reasonable given the relatively low volume and high cost of data at the time. However, retrospective analysis reveals the profound impracticality of such a tax in today’s digital age. A 1990s-era bit tax, if adjusted for inflation and the exponential growth of data, would amount to an astronomical $80 per gigabyte (GB) today. This figure underscores how rapidly technological advancement can render a seemingly benign tax prohibitive.
Senator Kelly’s proposed bit tax, set at $0.11 per GB, appears minuscule in comparison. For instance, an hour of interaction with an AI platform might consume a mere one to five megabytes (0.001 to 0.005 GB), leading to an initial tax burden of fractions of a cent. In its inaugural phase, the tax is projected to generate limited revenue, with an estimated $23.2 million in 2027. This estimate assumes that while individual consumers would likely fall below a proposed 10 GB exemption, businesses, particularly those leveraging AI APIs, would often exceed this threshold, making them the primary taxpayers. However, historical parallels with internet traffic, which doubled annually for about a decade in the early 2000s, raise significant concerns. If AI usage follows a similar trajectory, and the cost per unit of AI processing continues to collapse due to technological advancements, a fixed per-unit bit tax could quickly become an unbearable burden on technological adoption and development within a decade or two, disproportionately affecting the very technologies it aims to fund and potentially hindering the widespread benefits of AI.
The Excess Profits Tax: A Rocky Federal History
The concept of an excess profits tax has an even longer and more contentious history at the federal level, often emerging during periods of perceived corporate profiteering from national crises or unique economic conditions. A prominent example is the 1980 Windfall Profits Tax (WPT) imposed on oil and gas producers during a period of surging energy prices. The WPT was intended to capture what policymakers considered unearned profits from rising commodity prices and redirect them for public benefit.
However, analyses of the 1980 WPT revealed significant unintended consequences. Studies by the Congressional Research Service, among others, found that the tax reduced domestic oil production by an estimated 1.2 to 8.0 percent and consequently increased the nation’s reliance on foreign oil by 3 to 13 percent between 1980 and 1988, when the tax was eventually repealed. Furthermore, the WPT drastically underperformed its revenue projections, collecting a mere $80 billion compared to an anticipated $393 billion. This historical lesson is particularly relevant to Senator Kelly’s proposal, which is designed with an exceptionally narrow base. Initial analyses suggest that in 2027, almost the entirety of the excess profits tax, projected at $6.5 billion, would be borne by a single company, making it highly susceptible to firm-specific financial engineering, restructuring, or market fluctuations. Such a narrow base inherently introduces significant economic distortions and vulnerability to avoidance, potentially failing to achieve its revenue goals and unfairly targeting specific successful enterprises.
Categorically Unsound: The Fundamental Flaws of the Proposed Taxes
Sound tax policy is generally characterized by principles of neutrality, simplicity, transparency, and stability. Critics, including organizations like the Tax Foundation, argue that Senator Kelly’s proposed AI taxes deviate significantly from these fundamental tenets, rendering them economically unsound and potentially counterproductive.
The Digital Advertising Tax: Distortions and Disincentives
As a gross receipts tax applied to a business input, the digital ad tax is inherently non-neutral. It risks distorting production decisions, encouraging vertical integration (where companies perform multiple stages of production internally to avoid multiple layers of tax), and potentially stifling competition within the digital advertising ecosystem. The phenomenon of "tax pyramiding," where the same economic activity is taxed repeatedly through the supply chain, is a significant concern, ultimately increasing costs for businesses and potentially consumers, while disproportionately affecting low-margin businesses that operate with tighter financial constraints.
Furthermore, a critical irony arises from the proposed digital ad tax in the context of AI’s impact on workers. Many academics and policymakers have highlighted concerns that AI models are built upon vast datasets, often derived from the uncompensated work of human creators, such as authors, artists, and journalists. There are also fears that AI chatbots and summarization tools could divert internet traffic away from original content sources, thereby reducing monetization opportunities for these creators, who often rely on digital advertising revenue to sustain their work. By imposing a tax on digital ads, the "Make AI Work for Americans Act" could inadvertently diminish the income streams of the very content creators whose livelihoods are already threatened by AI’s rise, effectively taxing those most vulnerable to AI’s disruptive impact.
The Bit Tax: An Impending Burden on Innovation
The structure of the bit tax, as a fixed charge per unit of data, mirrors the flaws of its 1990s predecessors. While initially appearing negligible, the exponential growth in data usage, particularly with the proliferation of AI applications across various sectors, would inevitably transform this modest levy into a substantial, potentially prohibitive, cost. If AI development and adoption parallel the internet’s growth trajectory—characterized by rapidly increasing usage and collapsing costs per unit—the bit tax would soon become a significant barrier to technological progress. It would distort investment away from data-intensive AI models and towards less efficient, data-sparing alternatives, potentially stifling the very technological advancements the bill seeks to address. Moreover, administering such a tax would introduce considerable complexity for both taxpayers and the IRS, demanding sophisticated mechanisms to police technical avoidance strategies, including the offshoring of data processing services to jurisdictions without such levies. While converting the bit tax into an ad valorem tax (a percentage of value, like Rep. Greg Casar’s proposed token tax) might mitigate the escalating burden, it would still introduce distortions by making AI usage more expensive relative to other inputs.
The Excess Profits Tax: A Narrow, Unstable, and Discouraging Levy
The excess profits tax, with its extremely narrow base, is perhaps the most precarious of the three proposals. The analysis indicating that almost all of its projected 2027 revenue would come from a single firm underscores its profound non-neutrality. Tax policy that singles out specific companies or industries based on arbitrary financial metrics (such as a profit-to-receipts ratio) is inherently unstable, unpredictable, and prone to significant economic distortions. Such a tax can be easily circumvented through corporate restructuring, strategic accounting adjustments, or financial engineering, ultimately yielding far less revenue than projected while creating economic inefficiencies and potentially discouraging investment in the targeted sector.
Crucially, this tax also suffers from an "intertemporally asymmetric treatment" of innovation. It imposes a high tax rate on the payoff to successful innovation (excess profits) while allowing the costs of that innovation (research and development, capital expenditures) to be deducted at a lower, standard corporate tax rate. This imbalance fundamentally discourages risky but potentially transformative innovation, as the potential rewards for pioneering new technologies are significantly diminished while the upfront costs remain substantial. In essence, it penalizes success in a sector critical for future economic growth and global competitiveness.
Beyond Kelly’s Proposals: Pathways to Sound Tax Reform
While Senator Kelly’s intention to fund new spending and address the societal implications of AI is commendable, the proposed AI taxes are largely considered inefficient, complex, and potentially counterproductive. The digital ad tax risks targeting revenues unrelated to AI and burdening content creators. The bit tax is unlikely to withstand the rapid evolution of technology and data consumption. The excess profits tax, with its extremely narrow base, is vulnerable to avoidance and could stifle innovation critical for economic advancement.
Lawmakers seeking to genuinely fund new government initiatives or address the nation’s unsustainable federal deficits and debt would be better served by pursuing broader, more economically neutral tax reforms. Options that adhere more closely to principles of sound tax policy—neutrality, simplicity, transparency, and stability—include:
- Broadening the Income Tax Base: Reducing exemptions, deductions, and credits in the individual and corporate income tax systems to allow for lower statutory rates while generating more revenue. This approach minimizes distortions by taxing a wider array of economic activity more uniformly.
- Reforming the Corporate Income Tax: Streamlining the








