To quote a summary of their main point:

> Tax policy must stop disincentivizing human labor over machine labor. Because only human labor is taxed, the scale is tipped toward substitution. A consumption tax on A.I. tokens may flip this script.

Are the authors unaware of the fact that you can run models locally? Increasingly strong models, at that.

Either they ban local models -- or the tax is unenforceable and merely becomes a hosted model penalty/disincentive. Hosted models aren't that much more powerful, and there are plenty of other reasons to run models locally.

In general, instead of new AI taxes and penalties, how about incentives for human labor?

non-paywall: https://www.nytimes.com/2026/08/14/opinion/ai-policy-tax-tec...