by Cameron Reid
As artificial intelligence (AI) companies continue to pervade various sectors of our economy, policymakers must grapple with the question of how to regulate and tax the burgeoning industry. Taxing AI companies presents a complex challenge, necessitating a careful balance between safeguarding labour and the risk of hindering innovation.
Marietje Schaake argues in The Financial Times (FT) that it is time to discuss whether taxes need to be imposed on AI companies to protect labour and society from expected costs and industrial disruption from the AI revolution. While regulation is needed in the industry, with the first set of EU regulations coming into force in 2025 with the rollout of the AI Act, Forbes argues that although it is ‘well-meaning’, imposing tax at this early stage would be premature and risks stifling development. Consequently, many arguments for and against taxation need to be explored to understand the deeply contested issue of technological regulation and its possible unwanted side effects.
With Next Move Strategy Consulting predicting the market value of the AI industry at $1.85 trillion U.S. dollars in 2030 it’s clear that taxing AI companies could provide an opportunity to generate substantial revenue for governments. As AI technologies become increasingly prevalent in business operations, introducing a Pigovian tax on AI companies to account for the negative externalities of AI could provide governments with much-needed funds to invest in infrastructure, education, healthcare, and other public services. However, Chand et al. (2020) highlight that determining fair and effective taxation criteria for AI companies is inherently challenging. AI technologies are often intangible and rapidly evolving, making it difficult to assess their value accurately for tax purposes. This could lead to disputes and compliance issues between businesses and tax authorities.
Indeed, taxing AI companies could help level the playing field between traditional businesses and those leveraging AI. By imposing taxes, governments can prevent unfair competition and ensure that all businesses contribute equitably to the tax base. The Guardian furthers this point, adding that by implementing an AI tax it could prevent a situation where wealth is further concentrated in the few hands of BigTech. Furthermore, despite the substantial initial investment, AI systems offer the advantage of scalability with minimal marginal costs and improved economies of scale when compared to traditional manufacturing. As such, the effective marginal cost of AI production is zero, giving rise to the potential to disrupt value chains and making us question how we commercialise and view intellectual property. This can provide a unique regulatory opportunity to safeguard access to future technological innovation, ensuring that even the poorest countries will have these technologies available. As such, taxation and regulation together can allow for the equal distribution of AI and the wealth it generates, but to ensure this there would need to be a radical change in intellectual property laws within the industry. Particular attention should be paid to the EU AI Act as it progresses through the regulatory motions of the European Parliament in Brussels throughout 2024.
However, as Forbes points out, excessive taxation of AI companies could stifle innovation and hinder technological advancement by creating an uncompetitive market in which research and development aren’t favourable. AI start-ups and small businesses may struggle to survive under the burden of heavy taxation, leading to a less dynamic and competitive marketplace. EU’s competition and digital chief, Margrethe Vestager, told the FT, that the Act will not harm innovation and conversely promote the development of new technologies by providing a clear set of rules on how GenAI models are built from the foundation. The EU believes regulation will foster trust in the market through the introduction of a recognisable set of guidelines, making investment attractive and less risky for those still weary.
With the proliferation of AI technologies, there are legitimate concerns about job displacement and unemployment. With McKinsey projecting that 30% of working hours in America could be affected by automation and up to 12 million displaced workers in the next six years, Bill Gates argues that taxing AI companies could help fund programs aimed at retraining displaced workers and supporting those affected by automation-induced job loss. The logic of this argument follows from the ‘bank tax’ imposed on EU banks post-2008 financial crisis. Imposing taxes on companies engaged in systemically risky behaviour ensures that the potential externalities are accounted for. The revenue collected from these levies are allocated to a dedicated fund designed to safeguard against any future disruptions within the industry. However, this type of taxation could be seen as premature and unnecessary as Callum Williams from The Economist argues that there is no need for concern with the AI labour market. Williams argues that companies that are more rapidly adopting AI technologies are hiring more staff than those slower to adapt, and in better-paid positions with greater satisfaction levels. Williams highlights these companies are seen to be ‘hoarding’ staff as the skilled labour pool dries up which will ultimately stay the case unless a deep recession hits and rebalances labour demand and supply.
David Evan Harris highlights that the EU AI Act implementing taxes on AI companies could incentivise ethical AI development and deployment. By imposing taxes on certain AI applications, governments can encourage companies to prioritise the ethical implications of their technologies through regulatory oversight and controls. Thereby promoting measures such as privacy protection, banning applications like cognitive behaviour manipulation and emotional recognition for workplace micromanaging. However, like any tax policy, taxing AI companies may create opportunities for tax avoidance and evasion. Companies could exploit loopholes in tax regulations or relocate their operations to jurisdictions with more favourable tax environments, undermining the intended revenue-generating goals of the tax. Therefore, to avoid this, regulators must consider how AI policy can synergize with new taxation in countries with a history of facilitating corporate tax havens. Furthermore, it runs the risk of following the United States footsteps of making a scapegoat out of ‘unregulated’ crypto companies with levies born out of punishment for hypotheticals and scaremongering.
Indeed, as Emmanuel Macron highlights, unilateral taxation of AI companies by individual countries could lead to distortions in global trade and competitiveness. Disparities in tax rates and policies across jurisdictions may create uneven playing fields and discourage cross-border investment and collaboration in AI research and development. Higher taxes on AI companies could drive up the cost of AI products and services, making them less accessible to smaller businesses and organisations with limited financial resources. This could exacerbate existing inequalities in accessing AI technologies and hinder the broader societal benefits that AI innovation can offer. However, tax policies can be crafted to incentivize certain types of AI innovation while discouraging potentially harmful practices. For example, governments can offer tax credits or deductions for research and development expenditures related to socially beneficial AI applications, such as healthcare or environmental conservation.
Cognisant of this debate, taxing artificial intelligence companies presents both opportunities and challenges for policymakers seeking to navigate the complexities of the digital economy. While taxation can generate revenue, promote ethical practices, and address societal concerns, it must be implemented thoughtfully to avoid stifling innovation, promoting tax avoidance, and exacerbating global inequalities. Striking the right balance between taxation and innovation incentives is crucial for fostering a sustainable and equitable AI ecosystem that benefits society. As the AI landscape continues to evolve, policymakers must remain vigilant in adapting tax policies to ensure that they reflect the evolving nature of technology and its impacts on society.
The views expressed in this article are the author’s own, and may not reflect the opinions of The St Andrews Economist.
Photo via ChatGPT & DALL*E