Where Should You Put an AI Disclosure in an Email?

Put the AI disclosure where the reader will see it before they act on the email: a short line right under the greeting for autonomous AI replies, or a signature line for AI-drafted messages a human sends. Magic Teams AI builds these disclosures into the send layer of the email systems we install, so the notice fires automatically on the right messages and never gets buried in a footer nobody reads. The footer is a fine second home for a “how we use AI” link, but it should never be the only place the disclosure lives.

Here’s the scene that makes founders finally take this seriously. A prospect replies to a warm, human-sounding email, asks a follow-up, and gets a slightly-off answer. They scroll down, spot a tiny gray footer that reads “generated with AI,” and now they’re not annoyed at the mistake. They’re annoyed that you hid it.

Placement is the whole game. The same seven words land as honest transparency at the top of an email and as a buried confession at the bottom.

This guide gives you the exact spots, the legal reasoning behind each, and copy you can paste today.

Where exactly should the AI disclosure go in an email?

Put it in one of three anchored spots, ranked by how front-and-center the AI’s role is: the intro line under the greeting, the signature block, or the footer. The more the AI is doing the talking, the higher the disclosure needs to sit.

The governing idea comes straight from the FTC. A disclosure has to be “clear and conspicuous,” which the agency defines as difficult to miss and easily understood by ordinary consumers. In an interactive medium like email, it also has to be unavoidable, meaning a reader shouldn’t have to click through or scroll to a separate page to find it (Public Citizen / CLP Blog).

Proximity matters too. The FTC’s guidance is that a disclosure should sit close to the claim it qualifies, because physical closeness makes people connect the two (Davis+Gilbert LLP).

Here’s the placement map we use on every install. We match each disclosure spot to how much of the message the AI actually authored.

Read it left to right. If a human wrote the email and AI only cleaned up grammar, a footer link is plenty. If an AI agent is holding the conversation on its own, the disclosure belongs in the very first line the reader sees.

Option 1: The intro line, right under the greeting

Use this when an AI is doing the talking without a human in the loop for that message. This is the highest-visibility spot and the one regulators care about most.

For an autonomous agent, the disclosure should open the first reply. Intercom’s Fin recommends an email intro like: “Thanks for reaching out! I’m [agent name], [company]’s AI Agent. I’m well trained to answer your query, but you can reply and ask for the team at any time” (Fin by Intercom).

Notice what that line does. It discloses, it sets expectations, and it offers a human off-ramp, all before the reader has invested any trust.

Option 2: The signature block

Use this when a human sends the email but AI drafted the bulk of it. The signature is where readers already look to figure out who they’re talking to, so a disclosure there reads as natural context.

A clean pattern is one line directly under the name: “Drafted with AI assistance, reviewed and sent by [Name].” It sits inside the identity block, close to the human who’s accountable for the message.

This spot works because it borrows the credibility of the signature. You’re not tacking on a legal warning. You’re telling the reader exactly who did what.

Option 3: The footer

Use the footer for the lightest-touch cases, where AI only helped edit or format, and as a secondary home for a “how we use AI” link in every other case. The footer is the wrong place for your only disclosure when AI wrote the message, because most readers never scroll there.

The two-part approach compliance teams favor: a short primary line near the top or in the signature, plus a secondary linked “how we use AI” page describing your tools, review steps, and a contact (Humble Help Studio). The footer is where that link belongs.

Here’s the same three-tier logic as a quick reference. This table matches the AI’s role to a placement and a reason.

AI’s role in the email	Where to put the disclosure	Why
Autonomous agent replies, no human review	Intro line, top of email	Highest visibility; reader decides before trusting the answer
Human sends, AI drafted most of it	Signature block	Sits with the accountable human; natural context
AI lightly edited a human-written email	Footer + “how we use AI” link	Low stakes; transparency without disruption
Bulk marketing or cold email, AI-generated	Signature or intro, plus footer link	Regulated content; needs conspicuous, unavoidable notice

What do the laws actually require about placement?

The two frameworks that matter, the EU AI Act and the FTC, both care less about the exact pixel and more about whether the disclosure is impossible to miss. And “impossible to miss” effectively rules out footer-only placement for AI-generated messages.

Under Article 50 of the EU AI Act, systems that interact directly with people must be designed so those people are informed they’re dealing with AI, and providers of generative AI must ensure AI-generated content is marked as such (EU AI Act, Article 50). The Act is explicit about timing: that information has to be provided “in a clear and distinguishable manner at the latest at the time of the first interaction or exposure,” which is a strong argument for the top of the message rather than the bottom.

Those transparency obligations became applicable on 2 August 2026 (Bratby Law).

The reach is broad. Cold email typically lands in the “limited risk” category where transparency duties apply, meaning AI that generates subject lines, first lines, or body copy from prospect data needs disclosure (Puzzle Inbox).

Penalties aren’t trivial. Breaches of the AI Act’s transparency rules can reach up to 15 million euros or 3% of total worldwide annual turnover, whichever is higher (Zyphe).

On the US side, the FTC’s “clear and conspicuous” standard does the heavy lifting. In 2023 the agency updated its Endorsement Guides to formalize that framework, and the guidance evaluates disclosures on prominence, placement, proximity, and understandability (Davis+Gilbert LLP).

Regulators weight these four factors when they judge whether a disclosure passes.

Placement and proximity score highest for a reason. A perfectly worded disclosure in the wrong spot fails the test, because the reader never associates it with the message it’s qualifying.

For a deeper breakdown of which rules apply to whom, see our companion piece on whether AI email disclosure laws apply to your business.

Does it matter which type of email it is?

Yes. The right placement shifts based on whether the email is transactional, a support reply, a marketing send, or cold outbound. The stakes and the reader’s expectations differ in each.

Transactional and internal emails are the lowest-risk. A footer link is usually enough because nobody is being persuaded or sold to.

Support replies from an AI agent are the highest-risk for the “am I talking to a bot?” problem, which is why the intro line wins there. Marketing and cold email sit in the middle legally but high on the trust scale, because that’s where a hidden AI disclosure reads as manipulation.

Different email types carry different disclosure risk, and placement should follow.

The pattern is clean. The more the AI acts on its own, and the more you’re asking the reader to trust or buy, the higher and more visible the disclosure needs to be.

For cold email specifically, a hidden disclosure is the worst of both worlds. You take the legal risk and lose the trust benefit. Our guide on whether it’s legal to send AI-generated emails to customers covers the outbound rules in detail.

The MAGIC placement rule (our framework)

We coined a simple test so nobody on a team has to guess. It’s called the MAGIC rule, and it answers “where does this disclosure go?” in one pass.

MAGIC stands for More Autonomy, Give It Cover. The more autonomously the AI acts, the higher and more conspicuous the disclosure must sit. Full autonomy earns a top-of-email intro line. A human in the loop earns a signature line. AI as a light editing tool earns a footer link.

Here’s the rule as a decision tree you can hand to anyone.

The beauty of MAGIC is that it’s role-based, not channel-based. It gives the same answer whether you’re in Gmail, Outlook, a help desk, or a marketing platform. Autonomy determines altitude.

“

Stop asking where the disclosure looks least intrusive. Ask how much the AI is deciding on its own. The more it decides, the higher the notice sits. Autonomy sets the altitude.

What should the disclosure actually say?

Keep it short, plain, and specific about the AI’s role. The FTC standard requires that ordinary readers understand it, so skip legalese and name what happened.

Match the wording to the placement tier. An autonomous-agent intro should identify the AI and offer a human. A signature line should credit the human who reviewed. A footer link should point to a real page.

Here are field-tested lines you can adapt.

Copy-paste disclosure lines by placement

• Intro line: 'Hi, I'm Ada, [Company]'s AI assistant. I can answer most questions here, and you can ask for a human anytime.'
• Signature: 'Drafted with AI assistance, reviewed and sent by [Name].'
• Signature (support): 'Sent by [Name] with help from our AI assistant.'
• Footer link: 'See how we use AI in our communications'
• Marketing: 'Parts of this email were created with AI. A human on our team reviewed it before sending.'
• Cold outbound: 'This message was drafted with AI and reviewed by [Name] at [Company].'

One caution on wording. “AI-assisted” and “AI-generated” are not interchangeable, and using the softer term to describe a fully-automated message is exactly the kind of gap regulators flag. If the AI wrote it, say the AI wrote it.

The distinction has legal weight too. Even the cold-email guidance notes that human-in-the-loop review may change how a message is classified, so your words should match what actually happened (Puzzle Inbox).

Language and placement should also stay consistent across channels. Consumers want AI content labeled across formats, at 84% for written content in Fractl’s 2026 study (Fractl), so aligning your email disclosure with your website and social conventions builds recognition.

Does disclosing AI hurt or help your response rates?

It helps far more than founders fear, and hiding it costs you when readers eventually find out. The trust data is lopsided.

In a 2026 Meltwater and YouGov study of nearly 9,900 adults across seven countries, 86% of consumers said brands should disclose when content was created using generative AI (Meltwater / YouGov via Daily Guardian). Transparency isn’t just tolerated. It’s expected.

There’s a gap between that expectation and reality. Fractl found that only 20% of organizations always disclose AI use, while 33% never disclose at all (Fractl). That gap is the opening, because doing what most companies skip is a cheap way to stand out.

The demand for written-content labeling is overwhelming across the board.

• 84% Want written AI content labeled
• 16% No preference / against

There’s even upside pricing. In Usercentrics’ 2026 State of Digital Trust report, 52% of consumers globally said they’d pay more for brands transparent about AI, at a 7% average premium (Usercentrics). In Germany that climbs to 73% at a 9% premium, the highest of any market surveyed.

Trust in AI is also fragile, which raises the stakes on getting placement right. In the same Usercentrics study, 52% of consumers now trust AI less than humans with their data, and 47% had already taken a revenue-affecting action, like canceling or switching, over AI data concerns in the prior six months (Usercentrics).

Here’s how the transparency premium splits by market.

Put those facts together. People expect disclosure, they’ll pay for it, and they’re already suspicious. A visible, honest disclosure is a trust-builder. A hidden one is a time bomb.

If you’re worried the disclosure will make your AI emails feel robotic, the fix is voice, not hiding. Our guide on keeping AI emails on-brand and in your voice walks through it.

How do you implement this without checking every email by hand?

Build the disclosure into the send layer so it fires automatically on the right messages, and never rely on people remembering to add it. Manual disclosure is where compliance quietly breaks.

The steps we run on an install:

That last step matters more than it looks. An audit log of what got disclosed, where, and when is your evidence if a regulator or a client ever asks. It also lets you review AI email accuracy at scale, which we cover in our piece on auditing and reviewing AI email replies.

The reason we build this into the system rather than the signature settings is drift. A signature-template disclosure gets copied, forked, and stripped over months until half your team’s emails have no notice at all. A send-layer rule can’t be forgotten.

Key takeaways

• Placement follows autonomy. Autonomous AI reply goes to the intro line at the top. Human-sent AI draft goes to the signature. AI-edited human email goes to a footer link. That’s the MAGIC rule: More Autonomy, Give It Cover.
• The legal test is “unavoidable.” The FTC requires disclosures to be difficult to miss and not hidden behind a click or a scroll, which effectively rules out footer-only placement for AI-written messages (CLP Blog).
• The EU AI Act is live. Article 50 transparency obligations apply from 2 August 2026, and require disclosure “at the latest at the time of the first interaction” (EU AI Act, Bratby Law).
• Disclosure builds trust, not friction. 86% of consumers say brands should disclose generative-AI content, and 52% globally will pay a premium for AI transparency (Meltwater/YouGov, Usercentrics).
• Automate it. Build disclosure into the send layer with an audit log so it never depends on someone remembering.

Frequently asked questions

Where is the single best place to put an AI disclosure in an email?

For most business email, put it in the intro line for autonomous AI replies and in the signature block for AI-drafted messages a human sends. Those two spots satisfy the FTC’s proximity and unavoidability requirements because the reader sees the notice before or as they act on the message. The footer is only appropriate as a secondary link or for lightly AI-edited emails.

Is a footer disclosure enough for legal compliance?

Not when the AI wrote the message. The FTC’s “clear and conspicuous” standard requires disclosures to be unavoidable, and footers are, by design, the part of an email people skip (CLP Blog). A footer works as a home for a “how we use AI” link or for low-stakes AI-edited emails, but pair it with a higher placement whenever AI authored the content.

What exactly should an AI email disclosure say?

Keep it short, plain, and specific about the AI’s role, because the FTC requires ordinary readers to understand it. For an autonomous agent: “I’m [name], [company]’s AI assistant, ask for a human anytime.” For a human-sent draft: “Drafted with AI assistance, reviewed and sent by [Name].” Avoid softening a fully-automated message with the phrase “AI-assisted.”

Does the EU AI Act require AI disclosure in emails?

Yes, in many cases. Article 50 requires that AI systems interacting with people be designed so those people know they’re dealing with AI, and that AI-generated content be marked (EU AI Act, Article 50). These obligations apply from 2 August 2026 and reach content made available to the public, including much cold and marketing email (Puzzle Inbox).

Do I need to disclose AI if a human reviewed and edited the email?

If AI drafted the substance and a human approved it, disclose it. A signature line like “Drafted with AI assistance, reviewed by [Name]” is the right tier. If a human wrote the email and AI only fixed grammar or formatting, a footer link is usually enough. The line to watch is authorship: who actually generated the ideas and words.

Will disclosing AI lower my email response rates?

The evidence points the other way. 86% of consumers say brands should disclose generative-AI content, and trust drops sharply when people discover AI was hidden (Meltwater/YouGov). A visible, honest disclosure, especially one that offers a human off-ramp, tends to improve engagement and reduces the “is this a bot?” friction that kills replies.

What are the penalties for getting AI disclosure wrong?

Under the EU AI Act, transparency breaches can reach up to 15 million euros or 3% of total worldwide annual turnover, whichever is higher (Zyphe). In the US, FTC endorsement and deception rules carry civil penalties, and each non-compliant message can count separately. Beyond fines, the reputational cost of a discovered hidden disclosure often outweighs the legal one.

How do I disclose AI in cold outbound email without killing the pitch?

Fold it into the signature or a brief line near the sign-off: “This message was drafted with AI and reviewed by [Name] at [Company].” It reads as professional transparency rather than a warning label. Hiding AI in cold email is the worst option, because you carry the legal risk of AI-generated content and lose the trust you’d otherwise earn.

Should the disclosure look the same across every channel?

Yes. Consistency helps recognition, and consumers already want AI content labeled across formats, at 84% for written content (Fractl). Use the same core phrasing and a consistent “how we use AI” link across email, website, and social so the disclosure becomes familiar rather than alarming.

Do most companies actually disclose AI use today?

No, and that’s the opportunity. Fractl found only 20% of organizations always disclose AI use while 33% never do (Fractl). Since a clear majority of consumers expect disclosure, a business that discloses well stands out against the pack rather than blending into it.

How do I make sure the disclosure appears on every email?

Move it from a human habit to a system rule. Build the disclosure into the send layer so it’s injected automatically based on how the email was generated, and keep an audit log of every send and its disclosure state. Signature-template disclosures drift and get stripped over time; a send-layer rule can’t be forgotten.

If you’re rolling AI into your customer email and want the disclosure placed correctly, fired automatically, and logged for the day someone asks, that’s exactly the kind of thing we wire into the send layer during a Magic Teams AI install, and it’s a short conversation to figure out whether your setup needs it.