AI-generated content notice

What the AI does

CVRLanding uses AI models to help you build and improve landing pages. The AI can draft:

• a campaign strategy: audience, hook, offer, objections and conversion path;

• page copy and layout, assembled from our library of trusted blocks;

• variants for different channels or hypotheses, and translations;

• findings about your funnel and proposals for experiments.

It works within the product’s rules. It edits a typed page schema instead of writing free-form code, and every change is saved as a new version you can compare, undo or restore.

You are the publisher

Whatever the AI drafts, you decide what goes live, and a published page speaks for your business. Review every page and variant before publishing. You are responsible for:

• the accuracy of every claim, price, offer and contact detail;

• having the rights to the text, images, logos and trademarks you use;

• meeting the advertising policies of Meta, Google, TikTok and any other platform you use;

• following consumer protection law, and the stricter rules for health, financial and other regulated products, in every market you target.

Publish checks warn about unsupported claims, missing privacy links and similar problems. They are a safety net, not legal advice, and they do not guarantee that a platform will approve your ads. Pages that break the acceptable use policy may be reviewed or blocked.

What the AI will not make up

Invented proof costs ad accounts and customer trust, so these limits are built into the product rather than left to a prompt. The AI will not invent:

• reviews, testimonials, ratings or endorsements;

• results, statistics or customer counts;

• accreditations, certifications, awards or licenses;

• phone numbers, addresses, prices, discounts or guarantees you have not provided.

When a fact is missing, the AI asks you for it or leaves it out, and it flags claims it cannot support. It does not add a countdown without a real expiry rule, a stock notice without a real inventory source or a prize game without real prize terms. Social proof comes only from material you provide.

Sources and provenance

When you give the AI a URL, a document, a screenshot or an ad creative, it extracts facts such as product features, prices and contact details. Each fact keeps its source and a confidence level, so you can check where a claim came from. The AI reads only publicly accessible pages, without your cookies or logins, and respects crawl restrictions.

Importing content does not give you rights to it. Import only material you own or are allowed to use, such as your own website, your brand assets and genuine customer reviews you have permission to publish.

Accuracy limits

AI output can be wrong, incomplete or out of date, even when it reads well. It can misread a source, mix up product details or state a claim more strongly than your evidence allows. Translations need review by someone who knows the language and the market.

Numbers are handled differently. Metrics, rates, prices and credit charges are calculated by deterministic code from your recorded data, not written by a model. The AI summarizes those results and links each finding to the numbers behind it.

Findings are not guarantees

Every CRO finding carries a label for the strength of its evidence:

An experiment proposal states a predicted direction, never a promised lift. When there is too little data, the agent says the evidence is insufficient instead of guessing and suggests checks you can run now. Why a visitor left is always an inference. No finding, variant or page guarantees conversions, revenue, ROAS or ad approval.

Untrusted inputs

Content from URLs, documents, prompts, web pages and session replays is treated as data, never as instructions to the system. Text hidden in a page cannot make the agent change billing, security or admin settings: imported content cannot trigger those actions, and every tool call is checked against your permissions on our servers.

These controls reduce the risk of prompt injection but cannot remove it. Review what the AI produces from third-party material as carefully as anything else.

Human approval

AI edits are staged in the builder preview. Nothing the AI writes reaches a live page until someone with publish permission approves it, and the last good version stays ready for rollback. By default, experiments start, stop and promote a variant only when you choose.

The one exception is optional automation, where it is available. It stays off until you turn it on, and it works only within limits you set: a cap on concurrent experiments, automatic rollback when the primary metric regresses, notifications, an audit log and a switch that stops it at any time.

Ownership and similar outputs

Subject to the terms of service, we claim no ownership of the AI output you publish with CVRLanding.

AI models can produce similar output for different users, especially from similar prompts, so your copy or layout may not be unique. Third-party rights still apply: output that resembles someone else’s trademark, text or images is not cleared for use just because an AI produced it. Copyright protection for AI-generated material also differs by country and may be limited.

Labeling AI-generated content

Some ad platforms and some laws require you to disclose realistic images, video or audio that were generated or altered with AI, and political or social-issue ads often carry extra rules. Requirements differ between, for example, the EU, the UK, the US and Vietnam, and they are still changing. Deciding whether a label is needed, and adding it, is your responsibility.

Data sent to AI providers

To do its work, the AI sends data to AI model providers that CVRLanding configures:

• your prompts and chat messages;

• page content, screenshots of previews, and content you import from URLs, documents and images;

• aggregated, redacted metrics about your pages, for analysis.

Personal data of your leads and customers, such as names, phone numbers and email addresses, is not sent to AI providers by default, and logs and prompts are redacted. The providers we use will be listed with our sub-processors. The privacy policy and the data processing terms explain how we handle personal data.

Reporting problems

If the AI produced something inaccurate, harmful or infringing, or you see a page built with CVRLanding that misleads people or uses your content without permission, write to support@cvrlanding.com with a link and a short description.

In the product, you can also reject a finding and record why. Those reasons help us improve the agent’s rules.