Skip to content

New York SAFE for Kids Act rules: age checks by January 2027

Filip LuchianencoUpdated 7 min read

Last updated: 2026-10-13

New York Attorney General Letitia James and Governor Kathy Hochul released the final SAFE for Kids Act rules on July 28, 2026. The rules are 13 NYCRR Part 700. They are adopted under General Business Law Article 45 (§§ 1500–1508), the Stop Addictive Feeds Exploitation (SAFE) for Kids Act, signed on June 20, 2024 as Chapter 120 of the Laws of 2024. They set which platforms are covered, how accurate and how certified an age check must be, how parental consent works, and what happens to age check data.

When does the New York SAFE for Kids Act take effect?

January 25, 2027. Section 700.11 sets the effective date at "the 180th day after publication in the State Register", and from the release: "The final rules will be published in the State Register on July 29, 2026. The SAFE for Kids Act goes into effect 180 days later, on January 25, 2027." Under section 5 of the chapter law, the act takes effect "on the one hundred eightieth day after the office of the attorney general shall promulgate rules". January 25, 2027 is the date on the Senate's page for each section, such as § 1501.

The proposed rules were released on September 15, 2025 and published in the State Register as I.D. No. LAW-39-25-00001-P.

Which apps does the SAFE for Kids Act cover?

The rules apply to a "covered operator" of an "addictive online platform", and an "online platform" includes a "mobile application" (§ 700.1). In an addictive feed, media "shared or generated by users" are recommended, selected or prioritized based on "information persistently associated with the user or the user's device" or on past interactions. Search results, direct messages, subscribed accounts and groups, and the next item in a series are carved out.

A platform is covered when "20 percent or more of time spent by monthly active users on an online platform is spent on addictive feeds measured over any one six-month period an online platform chooses to measure in the prior calendar year" (§ 700.1(d)). A monthly active user is someone who reached an addictive feed and stayed on the platform "for at least one minute" (§ 700.1(y)).

An "Exempt Online Platform" has "fewer than 5 million monthly active users or fewer than 20,000 monthly active users who are covered minors", with an exception: platforms "whose primary user base is minors are not exempt" (§ 700.1(q)). A platform that loses the exemption has 180 days to comply the first time and 30 days on any later occasion (§ 700.2(c)).

The duties attach to users in New York. To determine where a user is, an operator must weigh "all reliable information" about location, including "technical information concerning a user's device" and data held "for marketing, content selection, or other commercial purposes", and look for users who conceal or misrepresent their location (§ 700.2(d)).

What do the rules require before an addictive feed or a nighttime notification?

An operator may serve an addictive feed only after an age assurance method determines the user is not a covered minor and the operator has no other actual knowledge that the user is a minor, or after verifiable parental consent (§ 700.2(a)). The same test applies to "Nighttime Notifications", defined in § 700.1(z) as notifications concerning an addictive feed "between the hours of 12 AM Eastern and 6 AM Eastern" (§ 700.3; statute § 1502).

A minor without consent keeps access to the platform "with a feed that does not include this feature" (§ 700.2(e)(4)), and the operator may not withhold features, degrade quality or raise prices other than as needed to comply (§ 700.9(b)). Consent starts with the minor: the operator must obtain the minor's valid consent before asking a parent, and the parent then goes through age assurance. Each parental consent method must include at least one option without a government ID and one without creating an account. For child-directed services and users under 13, the COPPA consent methods in 16 C.F.R. § 312.5 may be used (§ 700.2(e)). The notice to the parent may not disclose the minor's content selections, interactions, settings or the identities of other users (§ 700.9(d)).

How accurate must an age check be under the SAFE for Kids rules?

Accuracy is measured as a false positive rate, the share of minors a method identifies as adults (§ 700.1(s)). The "Accuracy Minimum" in § 700.1(b) sets these limits, plus "a rate of detecting method circumvention for an age assurance method that meets or exceeds 98%":

Age of the minor Maximum false positive rate
0 to 7 0.1%
8 to 13 1%
14 to 15 2%
16 8%
17 15%

The "Total Accuracy Minimum" (§ 700.1(gg)) applies the same limits to false positives and inconclusive outcomes combined. Every method an operator offers must be certified to the accuracy minimum, and at least one to the total accuracy minimum (§ 700.4(b)).

Certification is annual (§ 700.5(a)), by an accredited third party, against "ISO/IEC 27566-1:2025" or "IEEE 2089.1, 2024 Standard for Online Age Verification" or an equivalent standard (§ 700.1(l)). Testing also covers false negatives for ages 18 to 30. Operators must keep test results and certifications "for no less than 10 years" (§ 700.5(d)).

An operator that takes government ID must accept IDs "from all U.S. and non-U.S. jurisdictions" and, with narrow exceptions, offer a method without ID (§ 700.4(c)). A user classified as a minor can appeal and gets a decision or a request for more information within 10 business days (§ 700.6(c)).

What counts as actual knowledge that a user is a minor?

Four sources count as actual knowledge under § 700.4(a), each enough on its own: a self-declaration of minor status, an age assurance result, the operator's "good faith determination based on other, reliable evidence", and this one:

A covered operator's possession or access to covered user data for marketing, content selection, or other commercial purposes that, if applied to an age assurance method the covered operator provides, would result in a determination of minor age status

Explaining the same wording in the proposed rules, the Attorney General gave this example: "the platform may use data sufficient to conclude a covered user's age or age status to sell targeted advertisements directed to that covered user" (NPRM, page 87). The limit is on the same page: "merely possessing data about a user without the ability to discern the user's age status from that data does not implicate this requirement." Once an operator has actual knowledge, it must change the user's status from adult to minor within 10 business days (§ 700.4(d)(1)).

What data limits apply to age check data?

Data collected to comply "shall be the minimum necessary", "shall not be used for any purpose other than to comply", must be encrypted in transit and at rest, and must be deleted immediately once it is no longer needed to comply (§ 700.7(a)). Operators must keep for at least 5 years a record of each check, its method, date and result, plus monthly counts per method (§ 700.7(b)); the proposal had set 10 years.

Section 700.7 regulates only data collected for compliance with Part 700 (§ 700.7(d)). Under § 700.7(e), an operator "must comply with all other applicable data protection and security laws", and "the law that is more protective of a covered minor's privacy and safety shall govern" in a conflict. New York's Child Data Protection Act, which the rules do not name, is one such law. Per the release, that act "is in effect" and bars online platforms from collecting, using, sharing or selling the personal data of anyone under 18 without informed consent unless doing so is "strictly necessary"; it applies to users "actually known by the operator" to be minors and to services primarily directed to minors (GBL § 899-ee).

What are the penalties under the SAFE for Kids Act?

The Attorney General enforces the act. GBL § 1508 authorizes an action to enjoin violations and "to obtain civil penalties of up to five thousand dollars per violation", plus restitution, disgorgement "including but not limited to the destruction of unlawfully obtained data", and damages. The section opens "No earlier than one hundred eighty days after the effective date of this article". Article 45 takes effect on January 25, 2027, so under the text of § 1508 the first day for an action is July 24, 2027. The duties in §§ 1501 and 1502 apply from January 25. No separate enforcement date appears in the release or the rules, and no court ruling on the clause was found. Under § 700.8, the same action covers violations of the rules. The rules "apply equally" to operators that rely on a third party, such as an age check vendor (§ 700.9(a)).

NetChoice, a trade association, stated on July 29, 2026 that the final rules "cement an unconstitutional censorship regime". A search of CourtListener's federal docket index on October 13, 2026 found no challenge to Article 45 or Part 700 filed since July 1, 2026; the index covers only federal dockets, and New York state court filings were not searched.

Related: the New Jersey Kids Code Act, which also treats ages attributed for advertising as actual knowledge, the state app store age signal laws, and the 2026 COPPA Rule amendments.

What does an app scan show for the SAFE for Kids rules?

Two provisions turn on data an app already handles: § 700.4(a)(3) on marketing and content selection data as actual knowledge of age, and § 700.2(d) on device and marketing data as evidence of location. Both apply only to a platform inside the 20 percent scope and above the exemption thresholds.

When I scan an app, I record the ad and analytics SDKs in the binary and whether the build references the platform age APIs, Google's Play Age Signals and Apple's Declared Age Range (explained in the Android age verification post and the iOS post). On a test device, I capture which identifiers, age or birth-year fields and interest segments were sent to which ad and analytics hosts, and whether each request was sent before or after consent. SDK presence means the code is in the build. Transmission is established only by a capture. Whether an operator applied that data to an age assurance method, the share of user time spent on the feed, and the monthly user count are facts in the operator's own records, outside any capture. What a scan checks and the limits of a scan are on the methodology page.

I run these investigations for lawyers and for researchers working with lawyers, on one app or a list of apps in bulk. Sign up for a CanITrustThat account to run your own research, let us run an investigation for you, or read the teardowns.

App privacy law, applied to real apps.

Posts on new laws, fines and studies, and teardowns of the apps those laws apply to.

How we use your address: privacy notice. Prefer a feed reader? Use the RSS feed.