Platforms Insights

ChatGPT ad review: why ads get rejected and how to fix them

Anna Voronina
Anna Voronina
ChatGPT ad review: why ads get rejected and how to fix them

It took us four submissions to get our first ChatGPT ad approved. By the final attempt, we had removed a “Be First” claim and the OpenAI logo. Both raised policy concerns, but we could not establish which correction ultimately resolved the rejection.

For advertisers, that uncertainty is a problem. You can spend several rounds rewriting perfectly acceptable copy while the actual issue lies in business verification, a restricted category, or a landing page that OpenAI cannot access.

Every unsuccessful resubmission delays the campaign without necessarily bringing you closer to approval.

The first task is to identify what failed. From there, you can determine what needs changing, who should fix it, and whether an appeal makes more sense than another submission.

This guide combines OpenAI's official troubleshooting guidance with what we learned launching ChatGPT Ads through Evido by Aitarget.

First, find out what was rejected

In OpenAI Ads Manager Beta, open the Ads tab and check the affected ad's Status. If it shows Not serving, hover over the status to inspect the reason.

An ad may be awaiting review, rejected, or approved but unable to serve. These situations require different actions. An active setting also does not necessarily mean that an ad has passed review.

OpenAI documents the distinction in its review and serving troubleshooting guide.

Use the decision message to determine where to investigate:

What you seeWhat to investigateNext action
Ad rejectedCreative, claims, or destinationCorrect the affected asset
Business verification failedAdvertiser identity or affiliationCheck business information
Domain not approvedDomain and advertiser relationshipVerify ownership or authorization
Category or country restrictionOffer and targeting eligibilityReassess the proposed campaign
Approved but not servingCampaign, billing, or delivery settingsInvestigate serving issues

These are troubleshooting paths, not official OpenAI rejection codes. The actual decision message takes priority.

Step 1. Check eligibility and advertiser identity before touching the copy

OpenAI's advertising policies define which categories can be advertised. An otherwise well-written ad cannot make an ineligible product eligible.

Start with the actual offer and intended delivery countries. Some regulated categories require individual advertiser approval. Others remain prohibited.

If the rejection concerns category eligibility, compare the advertised product with the applicable policy and any approval granted to that advertiser. Do not assume an authorization for one service extends to every service the company sells.

Our guide to OpenAI ad policies covers the category restrictions and borderline cases in detail.

Advertiser identity is another potential issue. OpenAI checks business information, affiliations, and the relationship between an advertiser and its website.

For example, an agency may submit an ad for a client whose website belongs to a separate legal entity. If the account's advertiser identity or authorization is unclear, the team may need to correct its business information or provide evidence of the relationship.

A different headline will not resolve a failed business verification.

Once these conditions are clear, investigate the submitted ad itself.

Step 2. Save the rejected version before making changes

Keep a copy of what OpenAI reviewed: the ad title, body text, image, advertiser name, destination URL, rejection message, and relevant account identifiers. Record the submission date and any previous changes.

This becomes important when several elements are modified between attempts.

Suppose an advertiser rewrites a performance claim, replaces the image, and changes the landing page. The next submission gets approved, but the team cannot determine which correction resolved the rejection.

When the reason is ambiguous, investigate each potential issue separately. Change one material element at a time where practical. If several clear violations exist, correct all of them before resubmitting.

A simple submission log should connect each rejection to the correction made and the subsequent outcome. For agencies, keep these records separately for each client and ad account.

Step 3. Test the exact landing page that was submitted

A website can work perfectly for customers and still fail ChatGPT ad review.

OpenAI uses a crawler called OAI-AdsBot to inspect destinations. It must be able to access the landing page. OpenAI also recommends allowing OAI-SearchBot, which helps its systems understand publicly available content.

A page loading successfully in your browser does not establish that OpenAI's crawler can reach it.

Ask the person responsible for the website to inspect these conditions:

  • Robots.txt and server access. Confirm that OAI-AdsBot can access the submitted path and that the server returns the expected content.
  • Firewall and CDN restrictions. Look for blocked crawler requests, IP restrictions, or automated bot-protection rules.
  • Authentication and challenges. Check whether the destination requires login, CAPTCHA completion, or JavaScript-based verification.
  • Redirects and location rules. Follow the full destination path and confirm that the crawler reaches the intended public page.
  • Rate limiting. Investigate HTTP 429 responses, particularly after a large batch of submissions.

For infrastructure teams that require IP allowlisting, OpenAI publishes crawler ranges at adsbot.json and searchbot.json.

Consider a travel campaign linking directly to a hotel booking experience. Customers can open the booking application, but the room information sits behind an authentication screen or cannot be read by an automated crawler.

Linking to a public page with accessible information about the rooms and booking conditions may resolve the destination problem. Customers can continue to the booking flow from there.

OpenAI recommends using directly reachable web pages rather than destinations that require an application, login, or unsupported redirect.

The technical requirements are documented in OpenAI's crawler guidance for advertisers.

Step 4. Fix the claim or asset responsible for the rejection

If eligibility and destination checks reveal no issue, inspect the creative against the live offer.

OpenAI prohibits misleading claims about pricing, performance, capabilities, comparisons, and affiliations. It also restricts unauthorized trademark use and ads that imitate ChatGPT's interface.

A claim needs to remain true on the landing page

Imagine a hotel campaign with this copy:

Before
Title: Book your next city break
Copy: Every stay includes a free airport transfer.

The booking page reveals that airport transfers are included only with selected suites.

The advertisement promises a benefit that other rooms do not include.

If the hotel confirms that the benefit applies to those suites, a more accurate version would be:

After
Title: Explore city hotel suites
Copy: Book a select suite with an included airport transfer.

The landing page must identify which suites qualify and explain the relevant terms.

The same principle applies to discounted prices, free trials, and product capabilities. Claims should accurately represent what a customer can purchase or use.

Brand assets need their own check

An ad can describe a legitimate product yet suggest an affiliation or endorsement that the advertiser does not have.

Using the OpenAI logo without authorization, imitating a ChatGPT response, or presenting an advertisement as an official recommendation can raise policy concerns.

Our experience with these issues is documented in Evido by Aitarget's first ChatGPT Ads campaign.

Before resubmitting, have someone outside the creative team examine the complete advertisement. They should be able to identify the advertiser and understand the offer without additional explanation.

Step 5. Choose between resubmission and an appeal

The next action depends on whether the rejection is justified.

Correct and resubmit. If you identify a genuine policy or technical issue, fix it and submit the affected ad again. This applies to unsupported claims, misleading assets, destination mismatches, and crawler restrictions.

Appeal the decision. If the original submission complies with the applicable requirements, use Request Review or the available appeal option in Ads Manager. Explain why the stated rejection reason does not apply and include supporting evidence.

For account-level decisions, OpenAI directs advertisers to the appeal link in the decision email. These decisions follow a different process from individual ad rejections.

According to OpenAI's onboarding and policy troubleshooting guide, Ads Support is the next route when no appeal option is available or an appeal has remained undecided for more than seven days.

The support address is ads-support@openai.com.

A useful request includes the account ID, affected ad, exact rejection message or code, destination URL, and time of the notice. Add relevant evidence, such as authorization documents, policy references, or technical access logs.

If the rejection involves several teams, assign one person to coordinate corrections and the next submission. This avoids sending an updated creative while the website or account issue remains unresolved.

Step 6. Make sure approval leads to delivery

An approved ad can still record zero impressions.

Before reporting another moderation issue, confirm that the account is active, billing is valid, and the campaign is within its schedule. Check campaign and ad group status, remaining budget, targeting eligibility, and any reported delivery restrictions.

For product campaigns, verify that the selected product set contains eligible products.

If a newly launched campaign has no confirmed delivery or obvious blocker, OpenAI recommends allowing at least 24 hours before reporting zero delivery. Clear billing, account, or review errors should be addressed immediately.

See OpenAI's Ads Manager troubleshooting guide for the full delivery checklist.

Once impressions begin, the focus shifts to performance. Our ChatGPT Ads reporting guide explains which metrics to monitor and how to interpret reporting delays.

What to do before the next submission

A well-prepared ChatGPT Ads campaign should have its eligibility, advertiser identity, claims, and destination checked before submission. When a rejection occurs, the decision message should determine the next action.

For teams launching through Evido by Aitarget, these checks can be part of campaign preparation. We provide ChatGPT Ads access through our own API integration and Aitarget platform interface, allowing advertisers to create and manage campaigns without using OpenAI Ads Manager directly. We also check creatives, landing pages, and domains before submission and handle appeals when ads are rejected.

If you are preparing a ChatGPT Ads campaign or have encountered a rejection, contact Evido by Aitarget with your offer, destination URL, target countries, and available review feedback. We will assess what needs to be corrected and help prepare the next submission.

Approval ultimately depends on OpenAI's requirements and review decision. No partner can guarantee it.

Last updated:
Anna Voronina

Anna Voronina

Marketing manager

Anna is a marketer with experience in content marketing, digital growth, and go-to-market strategy. Over the years, she has worked with SaaS, media, and product-focused companies, developing content marketing, SEO, paid acquisition, brand communications, and marketing campaign launches. She is particularly interested in translating complex products and market specifics into clear messaging that helps businesses achieve their goals.

Help us improve our content

Is this article useful to you?

Tell us what you think

Ready To Optimize Your Campaigns?

Scale your performance marketing with automated solutions for Meta, TikTok, and Telegram.

Start Your Campaign