AI Daily Grind // AI news, freshly ground daily

← Slow Brew

Jobs & AI · ⏱️ 5 min

Robot tax: should companies pay when AI takes your job?

A fee when machines replace workers. A good idea. But who checks, and who cheats?

🎧 Listen as a podcast: Spotify Apple Podcasts

📖 What it is

Imagine your job goes to a machine. Your salary stops. Your payroll taxes stop too, and those taxes help pay for pensions and health care. So who fills the hole?

One answer is a "robot tax", also called an "automation levy". The idea is simple: when a company replaces people with machines or AI, it pays something back. (The robot does not pay: the people who deploy it do.) The money goes to retraining, or to the workers who lost their jobs.

The idea is not new. Bill Gates talked about it in 2017. The same year, the European Parliament voted against it. South Korea cut a tax break for automation, and some people called it the first robot tax. But it was not really a tax on robots.

Now the idea is back, because of AI. It comes in three main versions: - A fee per job replaced. You count the jobs that machines took, and the company pays. - A tax on AI use. For example, a tax on tokens (the small pieces of text or data an AI model reads and writes), or on AI revenue. - A tax on data centers, the big buildings full of computers that run AI. One idea is to tax the energy they use.

None of these is national law in the US. But one small place already started. In San Mateo County, California, a new rule took effect on October 29, 2026. It covers AI-controlled robots at work, in the areas outside the cities. If a robot takes a job, the business must choose: give the worker a job with equal pay, give notice and severance, or pay a fee into a retraining fund.

🔎 The facts

🗣️ My take

Bon, I will be honest with you. I do not know yet where I stand on this one. That is exactly why I wanted to write about it.

I like the idea. If a company saves money by replacing people, it is fair to ask it to help those people. And I think we need to do something soon. My feeling (not a study, just my feeling) is that in five years, maybe ten, a big part of today's workforce will not be doing the same work.

But I have a big doubt: how do you check it? A company can say, "We did not replace anyone, we just did not hire." It can call its AI a "tool", not a "worker". The tax experts say it too: first you must define what you tax. With AI, that is very hard.

It gets even harder across borders. If my country taxes AI and another country does not, the work can move there. Some places have weak rules, or rules nobody enforces. The countries with the tax could end up with more red tape and less economy, while others run free. That scares me.

So if I had to choose today, I would mix two things. Start with what is easy to see: data centers and AI computing are big, physical, and easier to count than "jobs replaced". Then build a shared rule between countries, maybe with independent auditors doing the checking, not the companies. What I would not do is skip the tax and just hope the safety net holds.

One more thing bothers me. OpenAI, one of the biggest AI builders, asked for robot taxes in its own paper. Why does an industry ask to be taxed? Is it generosity? Or do they know something about our jobs that we don't?

Why it matters

This is about your paycheque, and about the taxes that pay for pensions and health care. Today, that money comes mostly from work. If AI does more of the work, the money must come from somewhere else.

And again, speed is the problem. AI can change a job in months. Tax laws take years. Deals between countries take even longer. The bills in Congress are still only bills. The one real rule covers one county, and physical robots, not the AI inside your office software.

What you can do now: - Ask your employer: when AI takes over a task, what happens to the people? Retraining? A new role? - Ask your elected officials which version they support: per job, per token, or per data center. And who would check it. - Watch San Mateo County. It is small, but it is a real test.

Voilà, this is the question I ask myself, and now I ask you: should a company pay when it replaces a worker with AI, and who should make sure it really pays? Tell me where you stand.

Sources

Checked on 2026-11-09. This explanation was accurate on that day. AI changes fast, so check the sources and the date before you rely on it.

Get The AI Daily Grind by email, every morning, in simple words.

💬 Comments