SB 7 defines “automated decision systems” (ADS) used in employment and creates a package of transparency, procedural, and enforcement requirements for employers in California. The bill requires written pre-use notices (with a 30-day lead time for new deployments and a firm deadline for existing systems), an employer-maintained inventory of ADS, specific content disclosures (including data sources, vendor identity, and any quotas), and a worker right to access and correct ADS data.
It also prohibits certain uses of ADS (for example, inferring protected characteristics or using customer ratings as the primary input), limits sole reliance on ADS for disciplinary or termination decisions, and requires a human review when an ADS is primarily relied upon.
At a Glance
What It Does
SB 7 obliges employers to notify affected workers before and after using ADS that materially affect employment outcomes, disclose vendor and data details, and give workers controlled access to the data driving discipline or termination. It bars ADS uses that would infer protected characteristics or retaliate against workers and imposes human-review requirements when ADS play a decisive role.
Who It Affects
The law applies statewide to employers as broadly defined (including state and local governments, labor contractors, and independent contractors providing services), workers who face algorithm-influenced hiring, scheduling, discipline, or termination decisions, and the vendors or entities that create or supply ADS. Collective bargaining agreements can explicitly waive the statute under tightly defined conditions.
Why It Matters
SB 7 creates California-specific guardrails for algorithmic management that go beyond generic privacy rules by tying transparency to employment actions and by giving the Labor Commissioner enforcement authority. Employers and HR/AI vendors will need operational controls, documentation, and notice workflows; labor representatives and compliance teams gain new procedural tools to challenge ADS-driven decisions.
More articles like this one.
A weekly email with all the latest developments on this topic.
What This Bill Actually Does
SB 7 starts by defining key concepts: what counts as an automated decision system, what output from such systems is covered, who is an employer or worker, and what qualifies as an employment-related decision. The definitions are broad — ADS covers machine-learning, statistical, and analytics-derived systems that generate classifications, scores, or recommendations used to assist or replace discretionary decisionmaking — and the law explicitly excludes routine security software and simple tools like calculators.
Notably, the bill covers both employees and independent contractors and treats labor contractors as employers for these rules.
Before an ADS is first used to make employment-related decisions (other than hiring), SB 7 requires an employer to give a plain-language written notice to workers who will foreseeably be directly affected — at least 30 days before deployment. Employers already using ADS at enactment have a fixed compliance date.
Notices must identify the types of decisions the ADS could affect, general categories and sources of worker input data, any known key parameters that disproportionately influence outputs, the creators or vendors of the ADS, and, if quotas are used, the quantified targets and possible adverse consequences for failing to meet them. Employers must maintain a current inventory of deployed ADS and provide notices in the same languages used for routine communications.When an ADS materially contributes to discipline, termination, or deactivation decisions, the bill creates two procedural layers.
First, it forbids employers from relying solely on ADS output for those decisions; second, if an employer “primarily” relies on ADS output, the employer must have a human reviewer compile and evaluate other relevant information (examples listed in the text include supervisory evaluations, personnel files, work product, peer reviews, and witness interviews). At the time the employer communicates a discipline or termination based on an ADS, the worker must receive a plain-language post-use notice explaining the use of the ADS, who to contact, and how to request the worker data.
Workers can request one copy every 12 months of the prior 12 months of their ADS data used in such decisions, and employers must provide that data in a way that anonymizes third-party personal information.SB 7 enforces these provisions through the Labor Commissioner, who can investigate, issue citations, seek temporary relief, and bring civil actions; affected workers can also seek injunctive relief, civil damages, and attorney fees in superior court. The statute imposes a nominal civil penalty for violations and preserves local ordinances that provide equal or greater protections.
The bill also creates carve-outs: a clear and explicit waiver in a collective bargaining agreement relieves parties from the statute, and federal contracting or regulatory obligations that require ADS use are not impeded. It also cross-references California privacy law so that entities subject to the California Consumer Privacy Act and forthcoming CPPA automated-decision rules remain accountable under those frameworks in addition to SB 7.
The Five Things You Need to Know
Employers must give a plain-language pre-use notice at least 30 days before deploying an ADS that will affect current workers, with an April 1, 2026 deadline for ADS already in use when the law takes effect.
The bill requires employers to disclose vendor identity, categories and sources of worker input data, any key parameters known to disproportionately influence ADS outputs, and quantified quotas and possible adverse actions tied to those quotas.
An employer may not rely solely on ADS to make discipline, termination, or deactivation decisions; if the employer primarily relies on ADS, a human reviewer must compile and consider other relevant information (e.g.
supervisory evaluations, personnel files, work product, peer reviews, witness interviews).
A worker can request one copy every 12 months of the prior 12 months of worker data primarily used by an ADS for a discipline/termination decision; employers must anonymize third-party personal information in the disclosure.
Enforcement is through the Labor Commissioner (investigation, citations, temporary relief) and the civil courts; the statute specifies a $500 civil penalty for violations and preserves local laws and certain collective bargaining or federal contracting exceptions.
Section-by-Section Breakdown
Every bill we cover gets an analysis of its key sections.
Scope and operative definitions for ADS and covered parties
This chapter fixes the statute’s reach: ADS is defined broadly to cover machine-learning, statistical, and analytics-derived systems that generate scores, classifications, or recommendations used to assist or replace human discretion. The definition excludes basic IT and security tools but explicitly covers systems that materially impact natural persons. Importantly, the law defines 'employer' expansively — including state and local agencies and labor contractors — and defines 'worker' to include independent contractors. That combination widens compliance obligations beyond traditional employer–employee relationships and pulls in gig-economy participants and public agencies.
Mandatory 30‑day notice, inventory, and contents of disclosure
Employers must provide a separate, plain-language notice at least 30 days before deploying an ADS that will directly affect foreseeable workers; if an employer was already using an ADS on enactment, the bill sets a concrete compliance deadline. Notices must be in the same language as routine workplace communications and delivered through accessible channels. The required content is specific: anticipated decision types affected, categories and collection sources of worker input data, any known parameters that disproportionately affect outputs, vendor identities, and detailed quota information where applicable. Practically, this forces employers to operationalize inventory management, multilingual distribution, and pre-deployment documentation — functions many HR and legal teams currently lack.
Substantive prohibitions, human-review rule, data access, and collection limits
This chapter lists forbidden ADS practices (for example, using ADS to infer protected characteristics or to target workers for retaliation) and limits data collection to disclosed purposes. The law bars sole reliance on ADS for discipline/termination and imposes a conditional human-review requirement when ADS outputs are the primary basis for an adverse action; it also enumerates the types of 'other information' a reviewer should consider. The statute restricts customer ratings from being the only or primary data source and creates a regulated pathway for workers to obtain the 12 months of ADS data used against them, with anonymization requirements to protect third parties. These provisions will require employers to adjust decision workflows, vendor contracts, and data-handling procedures.
Immediate notification when ADS played a primary role in discipline/termination
When an employer primarily relied on ADS to make a disciplinary, termination, or deactivation decision, the employer must give the affected worker a stand-alone, plain-language post-use notice at the time the decision is communicated. That notice must identify a contact for further information and remind the worker of the right to request their ADS data. This creates a new HR touchpoint tied to adverse actions that could trigger follow-up requests, appeals, or investigations, and it requires employers to ensure decision timelines allow for immediate delivery of this notice.
Labor Commissioner enforcement, penalties, and carve-outs
Enforcement authority rests primarily with the Labor Commissioner, who may investigate, issue citations, order temporary relief, and pursue civil actions; aggrieved parties can also seek injunctive relief and damages in superior court with attorney’s fees. The statute sets a $500 civil penalty for violations and explicitly preserves stronger local ordinances. There are important exceptions: a collectively bargained waiver is valid only if it is clear, explicit, and includes protections from algorithmic management; regulatory or contractual federal obligations that require ADS use are not impeded; and entities subject to California’s consumer privacy law remain subject to forthcoming CPPA automated-decision rules. Employers will need to reassess collective bargaining language, federal contract clauses, and privacy compliance portfolios.
This bill is one of many.
Codify tracks hundreds of bills on Employment across all five countries.
Explore Employment in Codify Search →Who Benefits and Who Bears the Cost
Every bill creates winners and losers. Here's who stands to gain and who bears the cost.
Who Benefits
- Workers subject to ADS-driven discipline or termination — they gain notice, the right to request the ADS data used against them (one 12‑month copy per year), and the protection of a human review when ADS are the primary basis for adverse action.
- Independent contractors and gig workers — because the statute’s 'worker' definition explicitly includes contractors, these nontraditional labor categories receive the same transparency and procedural protections as employees.
- Labor unions and authorized representatives — the pre-use and post-use notice requirements, plus access and anti-retaliation protections, strengthen collective bargaining leverage and give authorized reps concrete information to negotiate or challenge ADS policies.
- Regulators and enforcement bodies — the Labor Commissioner gets clear jurisdiction and investigatory tools, including temporary relief and citation authority, which improves the capacity to address algorithmic harm in workplaces.
Who Bears the Cost
- Employers that deploy ADS (warehousing, rideshare, gig platforms, retail, call centers, and public agencies) — they must create inventories, translate notices, change operational workflows, document parameters and quotas, and support data-access requests.
- Third-party ADS vendors and HR-platform providers — they will face increased disclosure demands (vendor identity, parameters) and contractual obligations to support human review and data exports, raising product-development and legal costs.
- Small businesses and resource-constrained public agencies — the compliance burden (notice systems, anonymization, human-review processes) may be disproportionately costly relative to their budgets and technical capacity.
- The Labor Commissioner and courts — investigation, enforcement, and potential litigation demand staffing and technical expertise to evaluate ADS systems and disputed algorithmic outputs.
Key Issues
The Core Tension
The bill confronts a classic policy trade-off: protecting workers’ procedural rights and transparency (which demands disclosure, oversight, and human review) versus preserving employers’ operational flexibility, proprietary algorithms, and the economic benefits of automated management; the more the law protects workers, the more it raises compliance costs and potential disclosure of business-sensitive information, and the statute leaves open how to reconcile those competing interests in practice.
SB 7 builds useful procedural protections, but it leaves several implementation risks and trade-offs unresolved. Key terms like 'primarily relied on' and 'key parameters known to disproportionately affect the output' are fact‑intensive and legally fuzzy; employers and courts will litigate over the line between 'assistance' and 'primary reliance', as well as what counts as a sufficiently disclosed parameter.
The one‑request‑per‑12‑months cap for obtaining ADS data balances administrative cost against worker transparency, but it may be too restrictive for workers challenging an adverse action or for repeated harms occurring within a year. The anonymization requirement protects third parties, but it may complicate a worker’s ability to meaningfully challenge complaints or witness statements that informed an ADS decision.
There are also tensions between transparency and legitimate business interests. Requiring disclosure of vendor identities, model parameters, and quantified quotas risks exposing trade secrets or exploitable operational details; employers will seek to narrow or redact such disclosures, inviting disputes.
The statutory civil penalty—$500 per violation—is small compared with compliance costs and potential litigation exposure, which raises questions about whether nominal penalties will deter noncompliance. Finally, the law’s carve-outs—particularly the collective bargaining waiver and federal-contracting exception—create patchwork coverage and could incentivize employers to rely on bargaining or federal justifications to avoid transparency obligations, shifting protections unevenly across workforces.
Try it yourself.
Ask a question in plain English, or pick a topic below. Results in seconds.