Skip to main content
    Kriv AI

    AI for marketing

    Generative AI for marketing: what to check before a draft goes out

    Generative AI can draft emails, ad copy and web pages in minutes. The rules that apply to the finished piece have not changed, so the checking step matters more.

    Generative AI for marketing is the use of a model that writes text or makes images to draft campaign material such as emails, ad copy and web pages. A draft is not a finished piece: NIST describes confabulation as "the production of confidently stated but erroneous or false content", Google says "It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing", and the FTC says its CAN-SPAM rules cover "all commercial messages". Sources: S1 to S5, read 17 November 2026.

    plain answer

    Where generative AI fits in marketing work

    Marketing teams use generative models to produce first drafts: subject lines, ad variants, product descriptions, page copy and image concepts. The model writes and a person decides. This page is about the second half, the decision, because that is where the rules in the sections below apply.

    Workflow tools that move approved content between systems are a different topic, covered on our guide to AI marketing automation.

    plain answer

    A fluent draft can still be wrong

    NIST's Generative AI Profile (NIST AI 600-1, July 2024) lists confabulation as a risk and defines it as "The production of confidently stated but erroneous or false content (known colloquially as "hallucinations" or "fabrications") by which users may be misled or deceived."

    In marketing this shows up as an invented product feature, a wrong date, a statistic with no source, or a claim about a competitor. Google's guidance on generative AI content says "It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing." The same profile lists information integrity as a separate risk: content that may not "distinguish fact from opinion or fiction or acknowledge uncertainties."

    plain answer

    Search engines look at value, not at the tool

    Google's page on generative AI content says that "using generative AI tools or other similar tools to generate many pages without adding value for users may violate" its spam policy on scaled content abuse. Producing many thin pages is the risk the page names, and the tool is not the point.

    Google also says the review applies to metadata such as title elements, meta descriptions, structured data and image alt text, which can appear in Search results. A person should read those fields as well as the body copy.

    Google's page shows no update date. We read it on 17 November 2026.

    plain answer

    Email and reviews: the law looks at what is sent, not how it was written

    The FTC's CAN-SPAM compliance guide says the law "covers all commercial messages". Its rules include "Don’t use false or misleading header information", "Don’t use deceptive subject lines" and "Tell recipients how to opt out of receiving future marketing email from you." A model that drafts the subject line does not change those duties. The guide also says that when you hire another company to handle email, "you can’t contract away your legal responsibility to comply with the law."

    On reviews, the FTC announced on August 14, 2024 a final rule on fake reviews and testimonials. The release says the rule covers reviews by someone who does not exist, "such as AI-generated fake reviews", and "prohibits businesses from creating or selling such reviews or testimonials." A review or testimonial written by a model and presented as a customer's own experience is what the rule describes.

    plain answer

    Who owns what the model produces

    The U.S. Copyright Office's January 2025 report on copyrightability concludes that "given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output." Whether a given asset is protected is a question for counsel.

    NIST's profile also names an intellectual property risk: "Eased production or replication of alleged copyrighted, trademarked, or licensed content without authorization." Check generated images and copy for look-alike material before they are published.

    plain answer

    Put a named reviewer between the draft and the send

    The sources above point to one control: a person reads and approves each piece before it is published or sent. A short checklist is enough to start: every factual claim has a source a person opened, every price, statistic or customer name is confirmed, the subject line and sender details are accurate, the opt-out is present, and no review or testimonial is presented as a customer's that the customer did not write.

    That is a working habit and does not replace advice from your counsel. Kriv AI is not a law firm and this page is not legal advice. Our AI policies guide covers writing the rules down, and our AI marketing automation page describes drafting with a human approving before anything goes out.

    Straight answers

    Frequently asked questions about Generative AI for marketing: what to check before a draft goes out

    Is AI-generated content allowed in Google Search?

    Google's page on generative AI content does not ban it. It says using generative AI tools "to generate many pages without adding value for users may violate" its scaled content abuse policy, and that all AI-generated content should be fact-checked before publishing. Source: S2, read 17 November 2026.

    Do email rules change if a model wrote the email?

    The FTC's CAN-SPAM guide says the law "covers all commercial messages" and does not mention how a message is written. Its rules on header information, subject lines and opt-out apply to what is sent. Source: S3, read 17 November 2026.

    Can I use AI to write customer reviews?

    The FTC's August 14, 2024 release on its fake reviews rule says it covers reviews by someone who does not exist, "such as AI-generated fake reviews", and prohibits businesses from creating or selling them. Source: S4, read 17 November 2026.

    Can I copyright marketing copy a model wrote?

    The U.S. Copyright Office concluded in January 2025 that prompts alone do not give users enough human control to make them the authors of the output. Ask your counsel about a specific asset. Source: S5, read 17 November 2026.

    What is the main risk of generative AI in marketing?

    NIST names confabulation, "confidently stated but erroneous or false content", as a risk of generative AI. A person should check claims before a draft is published. Source: S1, read 17 November 2026.

    Talk to the team that would do the work

    Bring your requirements to a working session with the person who'll actually deliver.

    Book a Discovery Call