California limits AI firings, worker surveillance and neural data collection

Where the rules draw the line

The laws target practices that are already familiar to some workers: systems that map employees’ movements, track bathroom breaks or assess emotional cues. Amazon warehouse workers have complained about timed bathroom breaks; Kaiser Permanente nurses have said automated systems assessed their tone of voice with patients.

California Federation of Labor Unions president Lorena Gonzalez, who helped labor leaders around the country develop rules, called the laws a turning point. She said California workers were showing the country they did not have to accept these conditions.

Colorado, Connecticut, Illinois and Texas have passed narrower laws on workplace AI, and other states are preparing bills, Gonzalez said. California’s broader package comes as major technology companies ramp up AI spending and carry out large layoffs.

The new rules also arrive amid growing worker resistance to monitoring. In June, Meta paused a program that tracked employees’ computer activity to train AI models. A month later, dozens of workers sued, alleging that the company’s AI tools selected people for layoffs based on disability-related working conditions or because they were on medical or parental leave.

Enforcement is the open question

Robin Feldman, founder and director of the AI Law and Innovation Institute at the University of California College of the Law, San Francisco, called the laws an important step for worker advocates. But workers cannot sue over violations; only the state can enforce the rules.

That gap may matter as much as the restrictions themselves. A law that workers cannot enforce directly depends on the state to spot violations and act on them.

The package also leaves employers and advocates arguing over what should be regulated. Gonzalez said unions are watching new AI products because a product being sold is a good sign it could be put to use. The California Federation plans to use the current momentum to revive a proposal requiring employers to disclose their use of AI. That bill did not pass the state Assembly’s budget committee this year.

Employers, meanwhile, are more interested in how to introduce AI responsibly into their systems, said Danielle Ochs, a partner at Ogletree Deakins in San Francisco. She said the laws provide real enforcement measures, but it is not yet clear how broadly they will change workplace practice. Ochs also argued that requiring ten separate approval steps for each tool does not match how companies operate.

Opponents warn that restrictions could also block useful systems, such as technology intended to keep truck drivers from falling asleep at the wheel.

The reach of the rules is still uncertain

Workers and lawmakers are also debating AI’s broader risks. OpenAI and Anthropic have called for a slowdown in AI development amid national discussions of safety concerns, including fears that AI could destroy humanity. At the same time, workers are speaking more openly about job loss and the effects of surveillance and algorithmic control, said Annette Bernhardt, senior technology policy adviser at the University of California, Berkeley Labor Center.

My read is that California has drawn a meaningful line around specific uses, but the harder test will be whether enforcement can keep pace with tools employers adopt. The laws do not require employers to disclose AI use across the board, and workers cannot bring their own cases. For now, the rules offer a foothold for labor advocates, not a complete account of where AI is being used at work.