Companion chatbot laws · what to count and keep · updated
Companion chatbot laws: what each state asks you to count and keep
Four US states now have laws for companion chatbots, AI that keeps up a relationship with a user across sessions. New York's has applied since 5 November 2025 and California's SB 243 since 1 January 2026; Washington's HB 2225 and Oregon's SB 1546 apply from 1 January 2027. Each asks for an AI notice on a schedule and a crisis protocol. Three ask for a number every year; California's begins “Beginning July 1, 2027, an operator shall annually report to the office” (Cal. Bus. & Prof. Code § 22603(a)) (read 9 October 2026), and counts the crisis referral notifications of the year before. So the first report counts 2026: the referrals your protocol is making now.
What each state asks, and from when
| State | Law | Applies | AI notice | Yearly number |
|---|---|---|---|---|
| New York | GBL Article 47 | since 5 November 2025 | at the start of an interaction (at most once a day), then at least every 3 hours | none |
| California | SB 243 | since 1 January 2026 | where users could be misled; for known minors, every 3 hours with a break reminder, until 31 December 2026 | crisis referral notifications in the year before, to the Office of Suicide Prevention, from 1 July 2027 (what to count) |
| California | SB 1119, Adam's Law | mostly from 1 July 2027 | replaces SB 243's rules for known minors from 1 January 2027 | audit summaries to the Attorney General; the first audit by 1 January 2029 |
| Washington | HB 2225 | from 1 January 2027 | at the start, then at least every 3 hours; for known minors, at least every hour | last year's crisis referral notifications, on your website and in your app |
| Oregon | SB 1546 | from 1 January 2027 | where users would think it is a person; for minors, a break reminder at least every 3 hours | referrals in the year before, on a public website by 31 December |
What to count, starting now
The yearly numbers are counts your own systems can keep the moment your protocol shows a referral. Keep them by day or by month, beside the protocol version in force that day. The notice duties are shown best by the same kind of total:
- crisis-service referral notifications your protocol showed (each time it referred someone to a crisis line)
- sessions: continuing interactions, as your app counts them
- sessions that began without the AI notice
- sessions that ran past 3 hours without the AI notice being shown again
- sessions of users you know are minors that ran past 1 hour without the AI notice being shown again
- sessions of users you know are minors that ran past 3 hours without a break reminder
And keep the documents: each version of your crisis protocol with the date it took effect, where its details were published, and the notice's text with each release that changed it.
Does a law apply to your app?
That question is answered free elsewhere: AI Laws by State has an “Am I Affected?” checker, and the Future of Privacy Forum compares the California, Oregon and Washington laws side by side. These pages start after that: what to count and keep once one does. Whether one applies to you is for you and your counsel.
What these pages never ask for
The records these laws ask for are counts, dates and documents. Nothing here asks for, reads or keeps a conversation, a message or anything about a user, and nothing here detects or classifies anything: it counts what your own protocol already did.
Other states
Fenwick's list of 7 October 2026 names about a dozen more states with companion-chatbot laws dated from 1 January 2027 to 1 July 2027, among them Colorado, Rhode Island, Hawaii and Georgia. Each is added here once its statute has been read. For the disclosure laws that reach every chatbot, companion or not (the EU AI Act's Article 50, California's BOT Act, Utah), see AI disclosure.
A hosted record of your crisis-referral counts and notice cadence, with the Oregon and Washington public pages kept current for you, is not built
It would take your daily totals (never a conversation or anything about a user), keep them in a hash-chained log with each protocol version, and keep California's yearly draft and the Oregon and Washington pages current.
What exists today: these pages, and the counts and documents you keep yourself. okgate's hash-chained log records the tool calls AI agents make, not your app's events, so it cannot keep this record today.
Sources, each read on 9 October 2026
- New York General Business Law § 1702 (Article 47)
- Cal. Bus. & Prof. Code § 22603 (SB 243), FindLaw's copy current as of 1 January 2026
- Washington Legislature: ESHB 2225 as signed, Chapter 168, Laws of 2026
- Oregon Legislature: Enrolled Senate Bill 1546 (SB 1546-B), 2026 Regular Session
- AI Laws by State: tracker and the free “Am I Affected?” checker
- Future of Privacy Forum: comparison chart of Oregon SB 1546, Washington HB 2225 and California SB 243 (April 2026)
- Fenwick: The chatbot compliance wave: California's new law is not the only one to watch (7 October 2026)
- 988 Suicide & Crisis Lifeline
- Find A Helpline: crisis lines by country