When AI runs the firm, who pays tax?

AI deployment is raising issues of governance and, by extension, of taxation. Autonomous actions by digital agents can be attributed to human supervision, where the existing legal framework provides adequate regulation. The issue becomes more complex when AI deployment reaches a level of autonomy where human supervision breaks down. At that point, a new legal structure may be required to ensure responsible action among AI agents. It is a matter of calibrating legal changes to the level of autonomy achieved in AI adoption, thereby retaining the advantage of being reactive and, hence, not a threat to innovation.

Taxation similarly has a two-stage dynamic with AI. So long as productivity gains of AI are attributable to human beneficiaries, the existing tax regime will suffice. If AI systems become sufficiently intelligent, they could hypothetically run enterprises, and a different form of taxation would be called for. This may be far into the future, but a more pressing adjustment could be needed regarding income-tax. AI has begun to displace humans in enterprises, with consequential effects on tax revenue. Here, implications are continuous and tied to the rate of AI adoption.

Impact of job displacement will show up both in terms of tax forgone and in higher transfer payments to manage shifts in overall employment. The fiscal stress can be mitigated through taxes on consumption where AI agents are embedded in productive activity. This does not affect the pace of AI adoption as much as taxing the underlying capital structures does. Here, too, a point can hypothetically be reached where AI agents become consumers of economic output, and consumption taxes may not serve the original intent. For now, though, consumption taxes will have to carry more fiscal weight. These policy implications possess an urgency that lawmakers may not be fully appreciating. Several countries have initiated the process of adjustment. But it favours prioritising advancement in the AI race over the effects on governance and government revenue.