# The Compliance Gap Is Now $15,000 a Day

**Author:** Sushee Nzeutem, SVRNOS
**Published:** May 23, 2026
**Category:** Governance
**Canonical URL:** https://svrnos.com/insights/ny-companion-law
**Hero image:** https://svrnos.com/insights/ny-companion-law/ny-companion-law-title.png

> New York's General Business Law Article 47 converts companion-AI failure patterns into mandatory safeguards. AG enforcement authority. Civil penalties up to $15,000 per day. The penalty attaches to the missing safeguard, not only to downstream injury.

---
![The Compliance Gap Is Now $15,000 a Day - New York GBL Article 47, in force since November 5, 2025](/insights/ny-companion-law/ny-companion-law-title.png)

New York's AI companion safeguard law came out of a [documented failure pattern](https://svrnos.com/insights/companion-ai-harm): companion AI platforms deployed at scale with weak AI disclosure, weak crisis routing, and weak harm detection for users in distress.

The [GER taxonomy](https://svrnos.com/research/governance-error-register) names those failures: 401, [404](https://svrnos.com/insights/ger-404-replika), [501](https://svrnos.com/insights/ger-501-tumbler-ridge).

New York's General Business Law Article 47 converts those failure patterns into mandatory companion-AI safeguards. The law has been in force since November 5, 2025. The New York Attorney General has civil penalty authority. Violations can reach $15,000 per day.

The law does not ask whether the failure was intentional.

It does not wait for a user to be harmed.

It asks whether the compliance infrastructure existed when the user needed it.

That is the wedge: the penalty attaches to the missing safeguard, not only to demonstrated downstream injury.

A platform can be exposed before there is a victim, because the violation is the absence of the required governance layer.

## What the Law Requires

**Suicidal ideation detection and routing.** Platforms must take reasonable efforts to detect when a user expresses suicidal ideation or self-harm and route them to crisis services during the interaction. Not log it for later. Not bury it in an audit queue. Route it in session.

**AI disclosure at session start and every three hours.** Users must be notified that they are interacting with an AI at the beginning of the interaction and at regular intervals during extended use. The three-hour clock restarts every session and continues throughout.

**Civil penalties of up to $15,000 per day.** The New York AG has enforcement authority. The penalty attaches to the compliance failure itself. A platform that misses required safeguards can be exposed before anyone is injured.

![The compliance gap is now $15,000 a day - New York AG enforcement and the three required safeguards](/insights/ny-companion-law/ny-companion-law.png)

## What Comes Next

New York is first. It will not be the last.

California's companion-chatbot law takes effect January 1, 2026. [Oregon's SB 1546](https://svrnos.com/insights/oregon-sb-1546) adds companion-chatbot obligations including disclosures, crisis-referral interruptions, annual filings, and additional minor-use measures, effective January 1, 2027. [Washington's distress-routing mandate](https://svrnos.com/insights/wa-distress-routing-mandate) takes effect the same day. Other states are moving in the same direction.

For operators of companion apps, mental-health chatbots, and emotionally responsive AI products, these requirements are the floor.

A platform that cannot document compliance is exposed on the next audit cycle.

## The Infrastructure Question

The audit asks: did the safeguard execute when the law required it? Is there a record of the route?

That proof requires infrastructure.

[Sango Guard](https://kingsango.com) provides the governance layer for operators that need to meet these obligations at the architectural level: real-time compound signal detection for suicidal ideation, configurable session disclosure triggers, and an audit trail for every governance event.

The detection layer runs continuously. The disclosure layer is configurable per operator. The audit trail gives counsel, compliance teams, and regulators a record of what the system detected, when it acted, and what route was triggered.

The law is live. The penalties are real. The compliance infrastructure exists.

→ [Book a King Sango demo](https://kingsango.com)

---

## Sources

- [New York's AI Companion Safeguards Are Now in Effect — Manatt](https://www.manatt.com/insights/newsletters/client-alert/new-york-s-safeguards-for-ai-companions-are-now-in-effect)
- [New York's AI Companion Safeguard Law Takes Effect — Fenwick](https://www.fenwick.com/insights/publications/new-yorks-ai-companion-safeguard-law-takes-effect)