No. 05
Your Boss's Algorithm Now Needs a Human: AI at Work, Explained
California's No Robo Bosses Act and Connecticut's new AI rules put limits on using algorithms to discipline, fire and monitor workers. Here's what changed, what it means for Washington workers and employers, and what to ask at your job.
The Tea the short version
- California's No Robo Bosses Act bans relying solely on an algorithm to discipline or fire someone, starting July 1, 2027.
- California also banned AI tools that try to read workers' emotions.
- In Connecticut, employers must now give written notice when AI makes decisions about your job.
- Washington has no equivalent yet. Here are the questions every worker and employer should ask now.
Imagine getting fired by a spreadsheet.
No conversation. No chance to explain that the "idle time" flagged on your record was you helping a customer in the parking lot. Just a score, a threshold, and a notice.
For a lot of workers, especially in warehouses, delivery and call centers, that's not a thought experiment. Algorithms already schedule shifts, track productivity and flag people for discipline.
This fall, two states drew new lines.
California: humans have to be in the loop
On September 30, 2026, California's governor signed a package of workplace technology laws, according to the National Law Review:
- SB 947, the "No Robo Bosses Act." Employers can't rely solely on an automated decision system to discipline or fire someone. A human has to corroborate the decision, and workers get notice afterward. Takes effect July 1, 2027.
- AB 1883. Employers can't use AI workplace surveillance tools to recognize or infer employees' emotional states or collect neural data, with limited safety and defense exceptions.
- AB 1331. No workplace surveillance in bathrooms. (Yes, they had to write that down.)
That emotion-reading ban is the one I want you to sit with. Somebody built tools to guess how workers feel from their faces and voices, and sold them to employers. California said no.
Connecticut: tell workers when AI is deciding
Starting October 1, 2026, Connecticut employers must give written notice when they use AI to make a decision that affects the terms, privileges or conditions of someone's employment. Using AI to discriminate is now explicitly treated as employment discrimination under state law, according to the Connecticut Attorney General.
That's simple and powerful: you have a right to know when a machine is part of the decision about your livelihood.
Where Washington stands
Washington doesn't have an equivalent law yet. Our existing anti-discrimination laws still apply no matter what tool an employer uses. If an algorithm discriminates, "the computer did it" isn't a defense. But there's no specific rule requiring notice, human review or limits on emotion-reading tech.
I bring this up a lot in policy rooms, because the people most affected by automated management are rarely the ones invited to write the rules. That's why I believe so deeply in participatory design: the people living with a system should help shape it.
For workers: five questions to ask
You can ask these at any job, in any state. Ask HR, your manager, or your union rep.
- Do we use any automated system to track productivity, schedule shifts, or flag people for discipline?
- If a system flags me, does a human review it before anything happens?
- Can I see the data the system has about me?
- How do I challenge a decision I think is wrong?
- Is any tool analyzing my voice, face or emotions?
Write down the answers and the date. If the answers are vague, that tells you something too.
For employers: build the human in now
If you run a business, even a small one, you may already be using AI in hiring or management without thinking of it that way: résumé screeners, scheduling apps, call-center scoring, productivity trackers.
Here's my advice, whether your state requires it or not:
- Inventory your tools. List anything that scores, ranks, flags or recommends decisions about people.
- Put a human on every consequential decision. Hiring, firing, discipline, pay. A person reviews, a person decides, a person can explain why.
- Tell people. A short, plain-language notice: what you use, what it does, and how to ask questions.
- Skip emotion AI. The science is shaky, the legal risk is growing, and your team will never trust it.
- Test for bias. Ask your vendor how they test their tool for unfair outcomes by race, gender, age and disability. If they can't answer, that's your answer.
Compliance is the floor, not the goal. Teams that trust how decisions get made stay longer and work better.
Pour me another: Has an algorithm ever made a decision about your job? How did you find out? Tell me through the contact form.
Sip slow, stay curious.D
Receipts 2 sources cited in this post
- National Law Review natlawreview.com
- Connecticut Attorney General portal.ct.gov
I am not a lawyer, and nothing here is legal advice. Facts were current when this was written, and every claim links to its source. Spot something that has changed? Tell me and I’ll update it.

