California's Automated Decisions Law: 07.01.28 Deadline for AI Employers

For AI teams and employers deploying automated decision systems or workplace surveillance tools, California's new law converts algorithmic tooling from a productivity decision into a governed system. The July 1, 2028 effective date provides roughly twenty-one months to inventory systems, map vendors that access employee data, and document when and how humans review technology-generated recommendations — work that mirrors themes in NYC Local Law 144, Colorado's AI Act, and the EU AI Act. Builders and buyers alike should treat procurement checklists, model documentation, and human-in-the-loop audit trails as core compliance infrastructure, not optional add-ons.

California's 2026 legislative session closed with one of the most consequential packages of employment-law changes in recent memory, reshaping how employers across the state use artificial intelligence, monitor workers, manage workforce reductions, administer leave, and structure compensation agreements. The National Law Review's October 6, 2026 analysis lays out a multi-year compliance calendar: most of the new requirements take effect on January 1, 2027, while the automated decision systems law does not become operative until July 1, 2028, and separate training, court-procedure, and Paid Family Leave provisions carry later implementation dates. That staggered rollout is the single most important planning fact for employers and their counsel, because it converts what could have been a single panic deadline into a sequenced, multi-phase compliance program.

“

For AI teams and employers deploying automated decision systems or workplace surveillance tools, California's new law converts algorithmic tooling from a productivity decision into a governed system.

The centerpiece of the package is Assembly Bill 1697, which revises California's limits on so-called "stay-or-pay" provisions. The law generally prohibits employment-related terms that require a worker to repay a debt or pay a penalty, fee, or cost when the relationship with the employer ends. The prohibition applies to contracts entered into on or after January 1, 2027, and — critically — it reaches not only agreements signed at the start of employment but also certain agreements reached later in the employment relationship. The urgency measure also bars liability under the prior version of the statute for the period between January 1, 2026 and the measure's effective date, a retroactive-liability fix that closes a window of exposure created when the earlier law took effect. The statute preserves narrowly drawn exceptions for qualifying government loan-repayment and recruitment programs, transferable educational credentials, and approved apprenticeships. For employers that have used training-repayment agreements or signing-bonus clawbacks as retention tools, this is a structural change: the economics of onboarding investments must be rebuilt around non-repayment retention incentives.

Technology sits at the center of the 2026 reforms. The analysis directs employers to inventory their automated decision systems and surveillance tools, identify vendors that access employee data, and document when and how human reviewers evaluate technology-generated recommendations. This reflects a broader regulatory convergence: California is aligning with the transparency and human-review themes already visible in New York City's Local Law 144, Colorado's AI Act, and the EU AI Act, but attaching them specifically to the employment relationship. The July 1, 2028 effective date for the automated decision law gives employers roughly twenty-one months to map their systems, but the practical burden — vendor diligence, data-flow mapping, and human-in-the-loop documentation — is substantial and cannot be deferred to the final quarter.