AI Chatbot Disclosure Regulations — March 2026
Jurisdiction: Ontario-based Canadian company serving US and Canadian churches. Products affected: PewSearch chatbot (/chat/[slug]), CWA chatbot, voice agent.
TL;DR — What We Must Do
- Chatbot must identify itself as AI at the start of every conversation (required by Maine, CA, NY, CO, TX, NJ + FTC guidance)
- Privacy policy must list AI providers as data processors (PIPEDA + CCPA)
- ToS must include AI accuracy disclaimer (FTC + all state laws)
- Crisis protocols required for pastoral-care chatbots (CA SB 243 + NY A222A)
- Never let chatbot imply it's a pastor/counselor (FTC + proposed GUARD Act)
Compliance Status by Property
| Property | Privacy Policy | Terms of Service | Chatbot UI | Status |
|---|---|---|---|---|
| churchwiseai.com | Has AI sections, names OpenAI/Anthropic/Google | Has sections 5/13/14/15 for AI | N/A (admin only) | COMPLIANT |
| illustratetheword.com | Names OpenAI | Has Section 8 AI Disclaimer | Chat widget (basic) | MOSTLY COMPLIANT |
| pewsearch.com | NO AI mention | NO AI section | NO AI disclosure | GAPS — being fixed 2026-03-31 |
Enacted Laws (as of March 2026)
Canada
- PIPEDA: No explicit "identify as AI" mandate, but requires transparency about data processing and automated decisions. OPC expects meaningful disclosure.
- Bill C-27 / AIDA: DEAD — died on Order Paper Jan 2025 when Parliament prorogued. Will not return in current form.
US Federal
- FTC Act: Prohibits deceptive practices. AI chatbot outputs = company representations. Must not misrepresent chatbot's nature. Enforcement increasing (Sep 2025 inquiry launched).
- GUARD Act (S.3062): Proposed Oct 2025 — would require AI disclosure at start + every 30 minutes. Not yet passed.
US State Laws (Enacted, In Force)
| State | Law | Effective | Requirement | Penalty |
|---|---|---|---|---|
| Maine | Title 10 §1500-DD | Sep 2025 | "Clear and conspicuous" AI disclosure | Up to $10M (AG) |
| California | SB 243 | Jan 2026 | Companion chatbot: disclose AI + crisis protocols | $1K/violation + attorney fees |
| New York | A222A | Nov 2025 | Disclose at start + every 3 hours; crisis protocols | $15K/day (AG) |
| Colorado | CAIA | Jun 2026 | All consumer AI must disclose AI interaction | TBD |
| Texas | TRAIGA HB 149 | Jan 2026 | Disclose AI in consumer interactions | TBD |
| New Jersey | Bot Disclosure | Active 2025 | Disclose if commercial intent | Civil penalties |
| Utah | HB 452 | May 2025 | Mental health chatbot disclosure | Civil penalties |
EU (Reference)
- AI Act Article 50: Users must be informed they're interacting with AI. Effective Aug 2026. Not primary jurisdiction but relevant if any EU congregants use the chatbot.
Key Legal Considerations for Churches
- No religious exemptions: None of the enacted AI laws exempt churches or nonprofits.
- Pastoral care = mental health adjacent: Church chatbots handling grief, depression, suicidal ideation functionally overlap with mental health chatbots (CA, NY, UT laws).
- Extraterritorial application: All US state laws apply based on where the USER is, not where the company is incorporated.
- FTC jurisdiction: Canadian company selling to US churches is within FTC reach for deceptive practices.
Standard Disclosure Language
Chat header (persistent):
[Bot icon] AI Assistant
First message: "Hi! I'm [Church Name]'s AI assistant. I'm here to help answer questions about the church, services, and how to connect. For pastoral care or urgent needs, I can connect you with a real person."
Periodic reminder (for long conversations): "[Reminder: You're chatting with an AI assistant, not a pastor or counselor. If you need to speak with someone, just ask.]"
Sources
See full research in session notes (2026-03-31 persona testing session). Key sources:
- OPC Privacy and AI Hub: priv.gc.ca
- FTC AI Compliance: ftc.gov/ai
- California SB 243: Perkins Coie, Skadden analyses
- New York A222A: Fenwick, Morrison Foerster analyses
- Maine §1500-DD: Verrill Law analysis
- Colorado CAIA: Hunton analysis
- Air Canada v. Moffatt (2024): AI liability precedent cited in CWA ToS