Online platforms required to provide information pertaining to algorithm use, design transparency and user choice required, civil penalties provided, and rulemaking authorized.

A bill for an act

relating to consumer protection; requiring certain online platforms to provide

information pertaining to algorithm use; requiring design transparency and user

choice; providing civil penalties; authorizing rulemaking; proposing coding for

new law in Minnesota Statutes, chapter 325M.

BE IT ENACTED BY THE LEGISLATURE OF THE STATE OF MINNESOTA:

Section 1.

new text begin

[325M.35] ONLINE PLATFORM METRICS.

new text end

new text begin

Subdivision 1.

new text end

new text begin

Definitions.

new text end

new text begin

(a) For purposes of this section, the following terms have

the meanings given.

new text end

new text begin

(b) "Accessible user interface" means an interface that requires minimal user interaction,

including but not limited to clicks, taps, or similar actions, for a user to input data, make a

choice, or take an action while using a covered online platform.

new text end

new text begin

(c) "Algorithmic recommender system" means a computational process used to determine

the selection, order, rank, relative prioritization, or relative prominence of items provided

to a user on an online platform, including search results, ranking, recommendations, display,

or another method of automated selection.

new text end

new text begin

(d) "Covered minor" means a user who a covered business knows or should know, based

on knowledge fairly implied under objective circumstances, is a minor.

new text end

new text begin

(e) "Covered business" means a sole proprietorship, limited liability company,

corporation, association, or other legal entity that owns, operates, controls, or provides a

covered online platform. Covered business includes a joint venture or partnership composed

of businesses in which each has at least a 40 percent interest in the joint venture or

partnership. Covered business does not include a federal, state, Tribal, or local government

entity in the ordinary course of operations.

new text end

new text begin

(f) "Covered online platform" means an online platform that:

new text end

new text begin

(1) conducts business in Minnesota; and

new text end

new text begin

(2) uses one or more algorithmic recommender systems to determine the selection, order,

rank, or relative prominence of items provided to a user in whole or in part based on the

user's personal data, unless the data is:

new text end

new text begin

(i) based on user-selected settings or technical information concerning the user's device;

or

new text end

new text begin

(ii) a search query, provided the query is not associated with the user in the online

platform's data storage and is only processed to convey items in direct response to the user's

search.

new text end

new text begin

(g) "Default" means a preselected option adopted by a covered online platform for a

specific service, product, or feature.

new text end

new text begin

(h) "Engagement" means a user interaction with items on a covered online platform.

Engagement includes clicks; taps; comments; reshares; watching; dwelling; indications of

approval or disapproval, including but not limited to likes, dislikes, upvotes, or downvotes;

or any other form of interaction.

new text end

new text begin

(i) "Engagement data" means information that a covered online platform collects about

engagement on the platform. Engagement data does not include user survey data.

new text end

new text begin

(j) "High-value data" means user-provided data or predictions from user survey data

made by a covered online platform.

new text end

new text begin

(k) "Holdout group" means a group of covered online platform users that are exempt

from the application of algorithmic recommender system design changes.

new text end

new text begin

(l) "Item" means media eligible for display by a recommender system. Item includes

but is not limited to an individual post, account, group, page, channel, product, advertisement,

text, image, video, or audio file.

new text end

new text begin

(m) "Long-term holdout assessment" means a process in which a covered online platform

maintains a holdout group for a duration of at least 12 months.

new text end

new text begin

(n) "Long-term user value" means outcomes that align with an individual user's deliberate

and forward-looking preferences or aspirations, as expressed to a covered online platform

through high-value data.

new text end

new text begin

(o) "Long-term user value metrics" means the metrics a covered online platform uses

to measure long-term user value.

new text end

new text begin

(p) "Online platform" means a website, online service, online application, or mobile

application.

new text end

new text begin

(q) "Personal data" means information, including derived data and unique identifiers,

that is linked or reasonably linkable, alone or in combination with other information, to an

identified or identifiable individual or a device that identifies or is linked or reasonably

linkable to an individual.

new text end

new text begin

(r) "User" means a user of a covered online platform who is located in Minnesota. User

does not include a covered online platform operator or a person acting as an agent of the

covered online platform operator.

new text end

new text begin

(s) "User-provided data" means the following categories of information collected by a

covered online platform:

new text end

new text begin

(1) information expressly and explicitly provided by the user, including user preferences,

settings, search queries, prompts, and any other information expressly and explicitly provided

by the user that is not engagement data;

new text end

new text begin

(2) user survey data;

new text end

new text begin

(3) indicators or ratings expressly and explicitly selected by the user that are not

engagement data; or

new text end

new text begin

(4) other categories of data or more specific definitions of the data in clauses (1) to (3),

as may be defined by the commissioner of commerce by rule.

new text end

new text begin

(t) "User survey data" means a user response to questions that a covered online platform

or a third party acting on the covered online platform's behalf poses to users.

new text end

new text begin

(u) "Weight" means the individual numeric setting that controls the output of a

recommender system at a high level across a covered online platform's user base, including

the relative contributions of different factors to an item's ranking.

new text end

new text begin

Subd. 2.

new text end

new text begin

Applicability.

new text end

new text begin

(a) The requirements under this section are in addition to and

do not limit or restrict the application of other Minnesota law, including statutes, regulations,

and common law. If this section conflicts with one or more other laws, the law that provides

the greatest protection to the consumer applies.

new text end

new text begin

(b) Nothing in this section should be construed in a manner inconsistent with the First

Amendment of the United States Constitution or United States Code, title 47, section 230.

new text end

new text begin

Subd. 3.

new text end

new text begin

Design transparency.

new text end

new text begin

(a) A covered online platform that deploys an algorithmic

recommender system must prominently and conspicuously provide on the covered online

platform's website, service, or application:

new text end

new text begin

(1) a list of each algorithmic recommender system the covered online platform uses;

new text end

new text begin

(2) a description of each input to each algorithmic recommender system and the data

source for each input; and

new text end

new text begin

(3) the weights used in each algorithmic recommender system, categorized into four

quartile groups according to each weight's relative importance in contributing to the system's

output.

new text end

new text begin

(b) The commissioner of commerce must adopt rules to further clarify the information

that must be disclosed under paragraph (a).

new text end

new text begin

(c) A covered online platform must annually disclose the high-level objectives, key

results, and performance metrics the covered online platform uses to evaluate product teams

responsible for algorithmic recommender system design.

new text end

new text begin

Subd. 4.

new text end

new text begin

User choice; defaults.

new text end

new text begin

(a) For all services, products, and features where a

covered online platform uses an algorithmic recommender system that uses personal data,

the algorithmic recommender system must be configured, by default, to maximize one or

more long-term user value metrics.

new text end

new text begin

(b) A covered online platform must provide an accessible user interface that enables a

user to expressly and unambiguously communicate the user's preferences regarding the

types of items recommended and the types of items blocked in the output generated by the

covered online platform's algorithmic recommender systems. A covered online platform

must take all reasonable steps to ensure that the output of the covered online platform's

algorithmic recommender systems is consistent with the user's preferences.

new text end

new text begin

(c) Unless doing so is necessary to comply with this section or rules adopted pursuant

to this section, a covered online platform is prohibited from withholding, degrading, lowering

the quality, or increasing the price of a product, service, or feature for a user as a result of

the user exercising a right established under this section, including the user selecting an

algorithmic recommender system option or expressed preference regarding the type of item

recommended or blocked.

new text end

new text begin

Subd. 5.

new text end

new text begin

Covered minors.

new text end

new text begin

An algorithmic recommender system that uses personal data

and is provided by a covered online platform to a covered minor must be configured by

default to maximize one or more long-term user value metrics applicable to minors.

new text end

new text begin

Subd. 6.

new text end

new text begin

Long-term assessments.

new text end

new text begin

(a) A covered online platform must maintain at least

one holdout group and make all changes to the design of an algorithmic recommender

system subject to a long-term holdout assessment, under rules adopted under paragraph (c).

new text end

new text begin

(b) On an annual basis, a covered online platform must make publicly available in an

easily accessible location a long-term holdout assessment disclosure that includes:

new text end

new text begin

(1) the covered online platform's long-term user value metrics;

new text end

new text begin

(2) the aggregate and anonymized measurements for each metric across the holdout

group or groups; and

new text end

new text begin

(3) the aggregate and anonymized measurements for each metric across the rest of the

covered online platform's user base.

new text end

new text begin

(c) The commissioner of commerce must adopt rules governing the operation of long-term

holdout assessments under this section. The rules must include:

new text end

new text begin

(1) requirements for constructing holdout groups when performing long-term holdout

assessments under this section;

new text end

new text begin

(2) requirements for long-term holdout assessment disclosures required under paragraph

(b); and

new text end

new text begin

(3) in the commissioner of commerce's discretion, an exemption from the long-term

holdout assessment requirements under this section if a covered online platform changes

an algorithmic recommender system's design in a manner that reduces or prevents direct

and immediate harm to users without increasing user engagement or revenue for the covered

business.

new text end

new text begin

(d) At least annually, a covered business operating a covered online platform must, at

the covered business's expense, obtain an independent audit of the long-term holdout

assessments on the covered business's platform and the long-term holdout assessment

disclosure. To comply with this paragraph:

new text end

new text begin

(1) the independent auditor preparing reports must follow inspection and consultation

practices designed to ensure the reports are comprehensive and accurate; and

new text end

new text begin

(2) the covered online platform must provide the independent auditor with (i) full and

complete cooperation, and (ii) access to information and operations required to ensure the

report is comprehensive and accurate.

new text end

new text begin

Subd. 7.

new text end

new text begin

Enforcement.

new text end

new text begin

(a) A covered business that violates this section or rules adopted

pursuant to this section commits an unfair and deceptive act that violates section 325D.44.

new text end

new text begin

(b) The attorney general may enforce this section under section 8.31.

new text end

new text begin

(c) A covered business that violates this section or rules adopted pursuant to this section

constitute an injury to a user that entitles the user to the remedies available under section

325D.45 and:

new text end

new text begin

(1) monetary damages of $5,000, adjusted annually to reflect an increase in the Consumer

Price Index, per user per violation or actual damages, whichever is greater;

new text end

new text begin

(2) for a reckless or knowing violation of this section, punitive damages of $7,000,

adjusted annually to reflect an increase in the Consumer Price Index, per violation or actual

damages, whichever is greater;

new text end

new text begin

(3) reasonable attorney fees and litigation costs; and

new text end

new text begin

(4) other relief, including injunctive or declaratory relief, as the state or federal court

deems appropriate.

new text end