SAN ANTONIO — U.S. Representative Greg Casar proposed taxing artificial intelligence technology based on AI-generated outputs measured in units called “tokens,” with revenue directed toward job creation and retraining programs. The proposal, outlined in an opinion piece, is not formal legislation and would require introduction as a bill and passage through Congress to become law.
Casar argues that current tax policies favor automation because wages paid to workers are taxed while AI systems are not. “Our tax system basically gives companies huge tax savings when they automate a job, because we currently tax wages but not AI. If you replace a worker with an AI-powered robot, you save on payroll日消息 taxes: That’s functionally a tax break. That’s wrong, and an AI tax would start to change it by leveling the playing field,” he said.
Under the proposal, companies would be taxed based on both the number of tokens and the computing power used to train and run AI models, a design intended to prevent firms from manipulating token counts. He said the tax should target large corporations expected to profit billions by laying off workers, apply to AI providers rather than consumers, and impose higher rates on corporate users than individuals.
Casar warned that AI could cause “Great Depression–level unemployment” and called for a mass job creation program modeled on the New Deal’s Works Progress Administration. He said revenue should fund both direct employment in sectors with labor shortages—such as elder care, child care, teaching, and home insulation—and retraining, stating, “Only funding training at a time of mass unemployment may end up being like offering swimming lessons to passengers of the Titanic.” His goal, he said, is “to harness the wealth created by AI to create jobs at the same pace AI eliminates them, so that AI does not raise our unemployment rate.” He suggested structuring the tax mechanism after the Universal Service Fund, which adjusts fees quarterly to meet funding targets.