CCPA/CPRA Compliance for AI-Generated Consumer Emails

California's Consumer Privacy Act requires transparency when AI agents contact consumers. MultiMail's disclosure infrastructure embeds required notices automatically in every outbound email.


Why this matters

CCPA and its 2020 amendment CPRA require businesses to disclose when automated systems make decisions that affect consumers, and to provide opt-out mechanisms for automated decision-making (ADM). When an AI agent sends promotional emails, account notices, or personalized outreach to California residents, that communication falls under ADM disclosure obligations. Most email APIs have no concept of AI authorship — they treat every message identically regardless of whether a human or an agent composed it. That leaves your legal team manually auditing outbound campaigns, inserting disclosures by hand, and hoping nothing slips through. At scale, that process breaks down: agents can send thousands of emails before a compliance review catches a missing disclosure.


How MultiMail solves this

MultiMail enforces AI authorship disclosure at the gateway. Every agent-originated message automatically carries the X-AI-Generated: true header, a signed X-MultiMail-Identity claim (ECDSA P-256, covering the operator and mailbox identity, oversight mode, capabilities, and the AI-generated flag), and a human-readable disclosure footer. The ai_disclosure setting is locked on per mailbox and cannot be disabled, so disclosure can never be silently dropped from an AI-sent message. For California-resident outreach you add the CCPA-specific language and your own opt-out link to the message body before sending; MultiMail guarantees the AI-authorship disclosure is always present. The gated_send oversight mode gives your compliance team a review queue for AI-drafted messages before they reach consumers. Every send is recorded in the audit log, and recipients you have opted out are managed via the suppression list — together giving you an auditable trail for CCPA data subject requests and regulatory inquiries.

1

Identify California consumers in your contact list

Query your CRM or data warehouse for contacts with California billing addresses or California-resident flags. For those recipients, compose the message body with the CCPA-compliant disclosure language and your opt-out link required under Cal. Civ. Code § 1798.120. California residency and the choice of disclosure copy stay in your system; MultiMail's role is to guarantee the AI-authorship disclosure is always attached.

2

AI disclosure is enforced on every outbound message

You do not toggle disclosure per send — it is gateway-enforced. MultiMail appends the X-AI-Generated: true header and a signed X-MultiMail-Identity claim (base64url payload plus an ECDSA P-256 signature over the operator and mailbox identity, oversight mode, capabilities, and AI-generated flag), verifiable against the public key at /.well-known/multimail-signing-key. A plain-text disclosure notice is also injected into the message footer.

3

Route through gated_send for compliance review

With the gated_send oversight mode enabled, AI-drafted messages enter a human-approval queue before delivery. Your compliance team sees the full message — body, headers, recipient metadata, and disclosure status — in the approval UI. They can approve, edit, or reject. Approval decisions are logged with timestamps and reviewer identity.

4

Deliver with embedded disclosures

Once approved (or immediately if you use monitored mode for lower-risk sends), MultiMail delivers the message with the signed identity header and footer disclosure intact. The opt-out link in your footer points to your own preference page; when a consumer exercises their CCPA opt-out right, your handler adds them to the MultiMail suppression list so they are excluded from future sends.

5

Log compliance events for audit trail

Every send and oversight decision is recorded in the MultiMail audit log, retrievable via GET /v1/audit-log: the message, the mailbox it was sent from, the oversight mode, and any approval or rejection event. Combined with your own record of which recipients you identified as California consumers and which disclosure copy you sent, this gives you the trail to respond to CCPA data subject requests or to demonstrate compliance to regulators.


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. Get MultiMail ready: read https://multimail.dev/llms.txt, connect the MCP server, create a free inbox, and set up a verified sender. 2. In Salesforce Marketing Cloud, use Journey Builder with a Data Extension entry source or Automation Studio query activity to identify California contacts scheduled for AI-assisted promotional emails, account notices, or personalized outreach. 3. For each qualifying contact, draft the message with the required CCPA/CPRA automated decision-making disclosure, the business-approved AI authorship language, and the company opt-out link; personalize only from approved Salesforce contact and campaign fields. 4. Send or schedule the email through MultiMail in gated_send mode so compliance can review every AI-drafted message before release. 5. To go live, ask me only for Salesforce Marketing Cloud access, the verified sending domain details, approved brand voice, disclosure wording, and opt-out URL.

What you get

Machine-readable disclosures in signed headers

The X-MultiMail-Identity header is an ECDSA P-256 signed claim carrying structured AI-authorship data, verifiable against the public key at /.well-known/multimail-signing-key. This means downstream systems — mail clients, compliance tools, regulators — can verify disclosure authenticity without relying on body text alone.

Suppression list keeps opt-outs out of future sends

You host the CPRA opt-out flow required under Cal. Civ. Code § 1798.120 and record opt-outs in your own list. MultiMail enforces its suppression list at send time — bounced and unsubscribed addresses are rejected — so once an opted-out address is on it, future sends to that recipient are blocked at the gateway.

Auditable send and oversight log

Every send and oversight decision is recorded in the MultiMail audit log (GET /v1/audit-log): the message, the sending mailbox, the oversight mode, and any approval or rejection. Combined with your own record of California-consumer targeting, this gives you the trail for CCPA data subject access requests.

Compliance review before delivery

The gated_send oversight mode gives your legal or compliance team a structured review queue. Reviewers see the full message including embedded disclosures before anything reaches a consumer — catching edge cases that automated checks miss.

AI-authorship disclosure is never missed

The AI-authorship disclosure — the X-AI-Generated header, the signed X-MultiMail-Identity claim, and the footer notice — is gateway-enforced on every agent-sent message, so no individual agent can forget it or turn it off. You manage the CCPA-specific opt-out copy in your message body; MultiMail guarantees the underlying AI-authorship signal is always present.


Recommended oversight mode

Recommended
gated_send
Consumer communications carrying legal disclosure requirements benefit from human review before delivery. Under gated_send, your AI agent can draft messages autonomously and read inbound responses without friction, but outbound sends queue for approval. This gives your compliance team a checkpoint on AI-authored content before it reaches California consumers — important during initial CCPA/CPRA implementation when disclosure language and targeting logic are still being validated. Once you've run sufficient volume through the approval queue and confirmed disclosures are appearing correctly, you can migrate high-confidence message types to monitored mode while keeping edge cases gated.

Common questions

Does CCPA actually require disclosure when an AI agent sends email?
CPRA (the 2020 amendment to CCPA) added explicit rights around automated decision-making under Cal. Civ. Code § 1798.185(a)(16), directing the California Privacy Protection Agency to issue regulations on ADM disclosure and opt-out. The CPPA's final ADM regulations (effective 2025) require businesses to disclose when automated systems are used to make decisions that have significant effects on consumers, including targeted marketing. AI agents sending personalized outreach to California residents fall within this scope.
What disclosure does MultiMail attach, and what do I add for California consumers?
MultiMail always attaches the AI-authorship disclosure to agent-sent mail: the X-AI-Generated header, the signed X-MultiMail-Identity claim, and a human-readable footer stating the email was composed by an automated AI system. The CCPA-specific pieces — the consumer's right to opt out of automated decision-making under California law and a link to your opt-out page — are copy you add to the message body for California recipients. MultiMail guarantees the AI-authorship disclosure is present; you control the CCPA wording.
How does MultiMail identify California consumers?
MultiMail doesn't detect California residency — you identify California consumers from your own data (billing address, account registration state, IP-based geolocation, etc.) and add the CCPA disclosure copy to those messages before sending. MultiMail's role is to enforce the AI-authorship disclosure on every agent-sent message. This separation keeps your residency logic in your system while MultiMail guarantees the underlying AI-disclosure mechanics.
How do I wire up opt-outs to keep future sends compliant?
The opt-out link in your footer points to your own preference page. When a consumer opts out there, record it in your CRM or opt-out store and skip that recipient on future sends. MultiMail also enforces its own suppression list at send time — addresses that bounce or unsubscribe are rejected (GET /v1/suppression lists them) — so opted-out recipients are blocked at the gateway in addition to your own filtering.
Does the signed identity header satisfy CCPA's disclosure requirements on its own?
No. The X-MultiMail-Identity header is machine-readable and supports downstream verification, but CCPA compliance requires consumer-visible disclosure. MultiMail provides both: the signed header for audit purposes and the plain-text AI-authorship footer for consumers. The CCPA-specific opt-out wording is copy you add to the body; your legal counsel should confirm the specific disclosure language meets your obligations — MultiMail's enforced AI-authorship disclosure is a foundation, not a legal opinion.
What happens if I forget the CCPA copy when sending to a California consumer?
The AI-authorship disclosure is still attached — that part is gateway-enforced and cannot be skipped — so the message carries X-AI-Generated, the signed X-MultiMail-Identity claim, and the footer notice. But the CCPA-specific opt-out language and link live in the body copy you control, so those would be missing. The responsibility for identifying California consumers and including the CCPA wording sits with you. For campaigns where California residency is uncertain, consider adding the CCPA notice to all recipients to avoid compliance gaps.
Where is the audit trail, and what does it cover?
Send events and oversight decisions are recorded in the MultiMail audit log, retrievable via GET /v1/audit-log: the message, the sending mailbox, the oversight mode, and any approval or rejection. For a durable compliance record geared to CCPA's enforcement window, pull the relevant entries via the API and retain them in your own system alongside your record of which recipients you identified as California consumers and the disclosure copy you sent.

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.