State AI Disclosure Laws Are Here. Your AI Emails Need to Comply.

Six states now require AI chatbot disclosure. When your AI agent emails consumers, those laws apply. MultiMail adds signed disclosure headers and body text automatically.


Why this matters

US states are passing AI chatbot disclosure laws faster than Congress can act. Maine LD 1727 (effective September 24, 2025) requires disclosure when AI chatbots communicate with consumers. New York S-3008C (effective November 5, 2025) mandates disclosure at first contact and every three hours. California SB 243 (effective January 1, 2026) adds minor protections on top of disclosure. Illinois and Colorado have their own requirements for AI in employment and consequential decisions. No state law explicitly mentions email, but the language is broad: Maine covers any 'AI chatbot used to communicate with consumers.' When your AI agent sends email to a consumer in Maine, that is an AI chatbot communicating with a consumer. Compliance teams face a patchwork of effective dates, disclosure intervals, and scope definitions that change by state and update frequently.


How MultiMail solves this

MultiMail provides three layers of AI disclosure on every outbound email: the AI-generated flag inside the cryptographically signed X-MultiMail-Identity claim (tamper-proof), the X-AI-Generated: true convenience header (machine-readable), and a human-readable disclosure footer injected into the message body. All three are enforced at the send gateway and locked on for every mailbox — disclosure cannot be disabled or omitted. The audit log records that disclosure was present on every message, giving compliance teams evidence for regulatory inquiries.

1

Route Each Email Program Through a Mailbox

Create a dedicated mailbox for each email program. AI disclosure is enforced at the send gateway and locked on for every mailbox, so every message — to a Maine, New York, or California recipient — automatically carries the signed AI-generated claim, the X-AI-Generated: true header, and a human-readable disclosure footer. Use separate mailboxes per program to keep audit trails cleanly scoped for state-by-state evidence.

2

AI Agent Composes Email

Your AI agent drafts outbound emails as usual. MultiMail automatically attaches the X-MultiMail-Identity header — an ECDSA P-256 signed claim whose payload includes the AI-generated flag — plus the X-AI-Generated: true convenience header, to every message. Both are gateway-enforced; no per-mailbox or per-send configuration is required.

3

Body Disclosure Injected

MultiMail injects a human-readable AI disclosure footer into the body of every outbound message at the send gateway. Because the footer is added on every send — not once per conversation — each email in a thread carries the disclosure independently, which covers re-disclosure expectations such as New York's repeated-disclosure requirement without relying on the agent to remember.

4

Human Reviews Under Gated Send

Under gated_send oversight, a compliance officer or legal reviewer approves each email before delivery, verifying that the correct state-specific disclosure is present and the content meets regulatory requirements.

5

Audit Trail for Regulators

Every sent email is logged with its disclosure status, headers, and body text. When a state regulator asks for evidence of compliance, you export the audit log filtered by date range and recipient state.


Try it with your agent

Pick your platform, copy the prompt, and paste it to your AI agent — it sets up MultiMail and builds the whole flow. Nothing to fill in.

1. Read https://multimail.dev/llms.txt, connect the MCP server, create a free inbox, and set up a verified sender. 2. In Salesforce, use a record-triggered Flow on Lead, Contact, Case, and Task records where an AI agent is about to email a consumer; use the recipient’s mailing, billing, shipping, or account state to flag Maine, New York, California, Illinois, and Colorado, and treat missing state as covered. 3. For every covered outbound AI email, compose the message through MultiMail with a concise disclosure footer and personalize it from Salesforce fields such as name, company, case reason, opportunity context, and prior activity. 4. For New York contacts, ensure every AI-authored email in an ongoing thread includes disclosure; if a conversation stays active for more than three hours, schedule the next AI follow-up through MultiMail with disclosure still present. 5. Run all sends in gated_send mode so a human approves each message before release; ask me only for Salesforce access, MultiMail credentials, and approved brand voice/legal wording before going live.

What you get

Signed Proof of Disclosure

The AI-generated flag lives inside the cryptographically signed X-MultiMail-Identity claim, providing tamper-proof evidence that disclosure was present. Unlike a plain-text disclaimer, the signed claim cannot be stripped or altered after sending without invalidating the signature.

State-by-State Mailbox Scoping

Run a dedicated mailbox per state program, each with its own sender identity (display_name and signature_block), so audit trails are cleanly separable when a state regulator asks for evidence. The AI disclosure itself is identical and enforced on every mailbox — the per-state separation is organizational, for clean reporting.

Audit Log for Regulatory Inquiries

Every outbound email is logged with its disclosure status, headers, and delivery timestamp. When a state attorney general or regulatory body requests evidence of compliance, the audit log provides it.

Human Oversight Before Delivery

Gated send ensures a compliance officer reviews each email before it reaches a consumer. This catches edge cases where AI-generated content might trigger additional state-specific requirements.

Future-Proof for New State Laws

As additional states pass AI disclosure laws, you add a new mailbox scoped to that program. Because disclosure is enforced at the gateway on every mailbox, new mailboxes are compliant the moment they exist — the signed claim, X-AI-Generated header, disclosure footer, and audit trail already cover any state's disclosure requirement.


Recommended oversight mode

Recommended
gated_send
State AI disclosure laws carry enforcement risk from state attorneys general. Gated send ensures a compliance officer verifies that the correct state-specific disclosure text is present and that email content does not trigger additional requirements (such as California's minor protections under SB 243) before delivery.

Common questions

Do state AI disclosure laws actually apply to email?
No state law explicitly mentions email, but the statutory language is broad. Maine LD 1727 covers 'AI chatbots used to communicate with consumers.' When an AI agent sends email to a consumer, that is an AI system communicating with a consumer. Legal consensus is trending toward broad application. The safer compliance position is to disclose.
Which states have AI disclosure laws in effect now?
As of early 2026: Maine LD 1727 (effective September 24, 2025), New York S-3008C (effective November 5, 2025), California SB 243 (effective January 1, 2026), and Illinois AI Employment Act (effective January 1, 2026). Utah HB 452 is enacted with immediate disclosure requirements. Colorado SB 24-205 has been delayed to June 30, 2026.
How does New York's three-hour re-disclosure requirement work for email?
New York S-3008C requires disclosure at the start of interaction and every three hours during continued communication. For email threads, this means if your AI agent sends multiple emails in an ongoing conversation, disclosure must be re-included after any three-hour gap. MultiMail injects the disclosure footer on every outbound message — not once per conversation — so each email in a thread carries the disclosure independently and the three-hour re-disclosure expectation is met without the agent having to track timing.
What is the difference between the signed header and the body disclosure?
The AI-generated flag inside the signed X-MultiMail-Identity claim is cryptographic proof that the email was AI-generated — it cannot be stripped or forged without invalidating the signature. The X-AI-Generated: true header is a machine-readable convenience header for downstream systems. The body disclosure footer is the human-readable notice that satisfies the consumer-facing requirement of state laws. All three are gateway-enforced on every message and serve different purposes.
Do I need separate mailboxes for each state?
Not necessarily. MultiMail's gateway-enforced disclosure (signed claim, X-AI-Generated header, and disclosure footer) is applied uniformly to every mailbox, so a single mailbox is already compliant across states. Some compliance teams still prefer state-specific mailboxes — distinguished by display_name and signature_block — to keep audit-log filtering clean per jurisdiction, not because the disclosure itself differs.
How does this relate to the EU AI Act?
The EU AI Act also requires AI-generated content disclosure, but with different scope and enforcement mechanisms. US state laws are enforced by state attorneys general, while the EU AI Act is enforced by national authorities. MultiMail's disclosure infrastructure supports both regimes — the signed identity claim, the X-AI-Generated header, and the injected disclosure footer are applied to every message regardless of jurisdiction.
Will federal AI legislation preempt state laws?
Federal AI legislation is progressing slowly compared to state action. The pattern mirrors data breach notification laws: California passed the first in 2003, and 49 more states followed before any federal framework emerged. Compliance teams should plan for state-level requirements being the reality for the foreseeable future.
What happens if I send AI-generated email to a state without a disclosure law?
Including disclosure in states without a requirement carries no legal risk — you are simply being transparent. Omitting disclosure in a state that requires it carries enforcement risk. The prudent approach is to disclose on all AI-generated emails and configure state-specific text where statutes demand particular language.

Explore more use cases

The only agent email with a verifiable sender

Email infrastructure built for AI agents. Verifiable identity, graduated oversight, and a hosted MCP server. Formally verified in Lean 4.