SB295 makes it unlawful in California to distribute or rely on a pricing algorithm that processes confidential competitor information when that algorithm is intended or reasonably expected to be used by two or more competitors to set prices or other commercial terms. The bill defines key terms (including “pricing algorithm,” “competitor data,” and “commercial term”), creates a civil enforcement regime led by the Attorney General and local prosecutors, and attaches per-violation penalties and equitable relief.
This matters because it translates antitrust concerns about tacit, algorithm-assisted coordination into a targeted statutory prohibition focused on the flow and use of competitively sensitive data. Vendors of pricing software, marketplaces that surface pricing recommendations, and businesses that buy or rely on third‑party pricing tools will face new compliance and documentation duties, while enforcers get a clearer statutory hook to litigate algorithmic coordination claims.
At a Glance
What It Does
The bill forbids distributing a pricing algorithm or using its recommendation to set prices or other commercial terms when the algorithm processes confidential competitor data and is intended or reasonably expected to be used by competitors to coordinate terms. It also bans using recommendations known to incorporate competitor data that another competitor used to set similar terms.
Who It Affects
SaaS vendors of pricing and revenue-management tools, online marketplaces and platforms that provide pricing recommendations, retailers and service providers that adopt third‑party pricing tools, data brokers, and state and local prosecutors charged with enforcing competition laws. Insurers and certain insurance advisory organizations are expressly exempt from the main prohibitions.
Why It Matters
SB295 establishes a statutory approach to algorithmic collusion rather than leaving the issue solely to common‑law antitrust suits. That changes litigation strategy, compliance obligations, and product design for companies that build, sell, or consume automated pricing systems, and it creates new enforcement levers for state actors.
More articles like this one.
A weekly email with all the latest developments on this topic.
What This Bill Actually Does
SB295 first fixes the battlefield by defining the language it will use: a “pricing algorithm” covers any computational process — including systems built with machine learning — that recommends or sets a price or other commercial term inside California. “Competitor data” means confidential, nonpublic, competitively sensitive information belonging to two or more firms in the same market. “Commercial term” is broad: it reaches price, availability, output, discounts, and even rental rates and occupancy levels.
The core prohibition has two strands. One targets distributors: a firm may not distribute a pricing algorithm, or provide recommendations derived from it, to two or more competitors if the distributor intends or reasonably expects those recipients to use the algorithm to set prices or other commercial terms and the algorithm processes competitor data.
The other targets users: a business may not set a price or commercial term based on a pricing algorithm’s recommendation if it knows or should know the algorithm uses competitor data and that a competitor used the same system to recommend or set a similar term. The bill uses a mix of subjective intent and objective “should know” standards to reach both bad‑faith facilitators and negligent adopters.SB295 builds in several operational mechanics.
It treats violations as repeatable events: each authorized user of a distributed algorithm, each recommendation provided in violation, and each calendar month a prohibited recommendation is used counts as a separate violation. A defendant can raise an affirmative defense by proving they exercised reasonable due diligence before relying on recommendations — for example, by obtaining written assurances from the distributor that the algorithm does not process competitor data.
There is a time‑based safe harbor: competitor data collected more than one year before use is excluded from the prohibition.Enforcement is civil and entrusted to the Attorney General, district attorneys, county counsels, or city attorneys. Remedies include restitution, punitive damages, injunctive relief, reasonable attorney fees, and civil penalties (with a statutory cap described in the penalties section).
SB295 also makes contracts that violate the chapter void, preserves existing antitrust law, and excludes from coverage credit‑scoring tools governed by the Fair Credit Reporting Act or provided by consumer credit reporting agencies. Crucially, insurers and specified insurance advisory organizations are exempted from the principal prohibitions and related penalty provisions.
The Five Things You Need to Know
The bill prohibits distributing a pricing algorithm to two or more competitors, or making recommendations from such an algorithm available to them, when the distributor intends or reasonably expects the tool to be used to set similar prices or commercial terms and the algorithm processes confidential competitor data.
It also bans using a pricing‑algorithm recommendation to set a term if the user knows or should know the algorithm incorporates competitor data and that a competitor already used it to set or recommend a similar term.
Enforcement is civil (AG, local prosecutors) and treats each authorized user, each recommendation, and each month of use as a separate violation; remedies include restitution, injunctive relief, attorney’s fees, punitive damages, and civil fines up to $25,000 per violation.
A user can assert an affirmative defense by showing they exercised reasonable due diligence before adopting a recommendation, which the statute explicitly contemplates can include written assurances from the distributor that the algorithm does not process competitor data.
There are narrow carve‑outs and limits: insurers and certain insurance advisory organizations are exempt from Sections 17372–17374; tools covered by the Fair Credit Reporting Act or provided by consumer credit reporting agencies are excluded; and competitor data collected more than one year before use is not covered.
Section-by-Section Breakdown
Every bill we cover gets an analysis of its key sections.
Short title
Names the chapter the 'California Preventing Algorithmic Collusion Act of 2025.' This is the formal label that courts, regulators, and practitioners will use when citing the statute; it signals the Legislature’s explicit policy focus on algorithmic coordination risks.
Definitions — pricing algorithm, competitor data, commercial term
Provides the operational vocabulary the rest of the chapter relies on. Pricing algorithm is defined broadly to include machine‑learning systems. Competitor data is limited to confidential, competitively sensitive information from two or more competitors in the same market. Commercial term is expansive — price, availability, output, discounts, rental rates, and occupancy are all covered. The section also defines distribution broadly to include selling, licensing, subscriptions, and any means of making a tool available, which captures modern SaaS and platform delivery models. Notably, insurers and certain insurance advisory organizations are declared exempt from the main prohibitions, an explicit carve‑out that will affect enforcement in insurance markets.
Prohibitions — distribution, recommendations, and use
Sets out the two primary prohibitions: (1) a distribution ban aimed at entities that provide pricing algorithms or recommendations to multiple competitors when the distributor knows or should know the algorithm processes competitor data and the distributor intends or reasonably expects the recipients to use it to set commercial terms; (2) a user ban preventing firms from adopting algorithmic recommendations when they know or should know the tool incorporates competitor data and has been used by a competitor to set similar terms. The section builds objective and subjective culpability elements into the prohibitions, and includes an affirmative defense for users who can show reasonable due diligence (with written assurances given as an example). It also defines how repeated conduct is counted and provides a one‑year lookback safe harbor for stale competitor data.
Enforcement and remedies
Authorizes civil enforcement by the Attorney General, county or city attorneys, and district attorneys. Remedies mix equitable and monetary relief: courts may order restitution, injunctive relief, reasonable attorney’s fees and costs, punitive damages, and civil penalties up to $25,000 per violation. The statute instructs courts to weigh factors such as seriousness, number and duration of violations, willfulness, defendant’s finances, and cooperation when setting penalties — a list that gives courts discretion to calibrate fines to the conduct and the defendant’s ability to pay.
Void contracts
Declares that contracts violating the chapter are void to that extent. Practically, this can invalidate distribution or licensing agreements that facilitate prohibited algorithmic coordination and creates a private‑law consequence in addition to public enforcement.
Preservation of antitrust laws
Affirms that the chapter does not limit or impair existing antitrust statutes. This signals the Legislature’s intent that SB295 supplements, rather than replaces, federal and state antitrust remedies and preserves concurrent enforcement paths.
Exemptions for credit tools
Exempts credit scores and other computational tools that are subject to the Fair Credit Reporting Act or provided by a commercial credit reporting agency, provided they are not used to facilitate direct coordination of commercial terms among competitors. This carve‑out prevents overlap with federal credit‑reporting regulation but narrows the exemption when such tools are used for coordination.
This bill is one of many.
Codify tracks hundreds of bills on Justice across all five countries.
Explore Justice 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
- Consumers and end customers — by reducing the risk that algorithmic coordination will lift prices, restrict availability, or standardize commercial terms across competing sellers in covered markets.
- Small and independent sellers — who are less able to monitor competitors’ algorithmic behaviors and therefore stand to benefit from a rule that limits back‑channel pricing coordination facilitated by shared tools or data.
- State and local prosecutors — who gain an express statutory cause of action focused on algorithmic coordination, with clear remedies and penalty factors to pursue enforcement.
- Renters and residential markets — because the definition of commercial term explicitly covers rental rates and occupancy levels, tenant markets affected by automated pricing may receive protective effects.
Who Bears the Cost
- SaaS vendors, pricing‑software firms, and marketplaces — which must redesign products, tighten data inputs, implement compliance controls, and document data provenance to avoid running afoul of the prohibitions.
- Businesses that rely on third‑party pricing recommendations — which face new legal risk and must perform due diligence or obtain written assurances, increasing procurement and legal costs.
- Data brokers and aggregators that supply competitively sensitive information — whose commercial models may be constrained if their datasets are treated as ‘competitor data’ and thereby restricted in downstream use.
- Corporate legal and compliance teams — which must build policies, auditing, and recordkeeping practices to prove due diligence and defend against claims where a regulator alleges 'should know' knowledge.
- State and local budgets and prosecutors — which may need additional technical resources and expertise to investigate complex algorithmic systems and litigate against software vendors.
Key Issues
The Core Tension
The bill’s central dilemma is familiar: protect markets from tacit, algorithm‑enabled coordination without unduly restricting legitimate, procompetitive uses of automated pricing and market intelligence. Tight rules deter collusive outcomes but also push vendors and users toward conservative design and heavy compliance — a trade‑off between preventing anti‑competitive outcomes and preserving the efficiencies of data‑driven pricing.
SB295 packs several hard implementation questions into a compact statute. First, the line between permissible market‑monitoring and prohibited processing of competitor data is fact‑heavy: firms routinely ingest public price postings, scraped listings, and aggregated market signals.
The bill’s definition of competitor data as 'confidential, nonpublic, competitively sensitive' will force litigants and regulators to parse datasets and prove what was truly nonpublic and sensitive — a costly and uncertain task. Second, the statute mixes intent standards with an objective 'should know' test; proving what a distributor 'reasonably expected' or a user 'should have known' about an algorithm’s inputs and downstream use will hinge on internal documentation, metadata, and corporate communications that are often opaque.
The affirmative defense — reasonable due diligence including written assurances — is practical but blunt. Written assurances may reduce liability exposure, yet they can be gamed and may not reflect actual data flows inside complex models.
The one‑year safe harbor for older competitor data raises questions about how the statute measures data age for derived features and models that continually retrain on historical inputs. Finally, the exemptions (notably for insurers and FCRA‑covered credit tools) risk regulatory arbitrage: firms may shift pricing or coordination activity into exempted channels or products, or narrow the definition of 'processing' to avoid coverage.
Altogether, the statute gives enforcers a workable tool, but it raises evidentiary and compliance frictions that could chill legitimate uses of dynamic pricing and increase transactional friction for lawful products.
Try it yourself.
Ask a question in plain English, or pick a topic below. Results in seconds.