If App Review flagged your AI chat app under Guideline 1.2 and the message points to user-generated content, the first instinct is usually to argue that the chat is not actually generated by users, it is produced by a model. That argument has not been working since February 2026. Apple now treats AI-mediated chat as the same surface as any other social feature, and the build either ships the four safeguards or it does not pass.
Short answer
On February 6, 2026, Apple updated the App Review Guidelines to clarify that random or anonymous chat is subject to Guideline 1.2 on user-generated content. AI chat apps are caught by the same rule whenever users can save, share, screenshot, or react to model output, or when one user's prompt influences another user's session. The four requirements are spelled out in the guideline itself: a method for filtering objectionable material, a mechanism to report offensive content, the ability to block abusive users, and published contact information. Twenty-four hours is the response window most developers see in the Resolution Center, although Apple does not publish a number.
What you should know
- The February 2026 update made the read explicit. Apps with random or anonymous chat now fall under Guideline 1.2, and AI chat apps that share or store output behave the same way to a reviewer.
- Four safeguards are required, not optional. Filtering, reporting, blocking, and a real support address. Missing any single one is enough to trigger a 1.2 rejection.
- The Report action belongs on every message. Reviewers test by long-pressing a recent AI reply and looking for the option in the action sheet.
- Blocking has to actually block. A toggle that hides messages locally while the counterparty keeps sending them does not pass.
- The 24-hour response window is community-tested, not officially published. Apple writes "timely responses"; the practical cadence is one business day.
Why does Apple class AI chat as user-generated content?
The cleanest read is the Apple Developer News post from February 6, 2026, which says directly that random or anonymous chat is subject to Guideline 1.2. The wording is about the chat experience, not about who or what is on the other side of the conversation. When the counterparty is a model, the user is still generating prompts, the model is still producing text that lands inside the user's session, and in most AI chat apps that text is shareable, copyable, or visible to other users in a group thread.
Reviewers do not draw a philosophical line between content the model produced and content the user produced. They look at the screen. If the screen contains text another user can see, or text the reporter can later regret having posted, the app is UGC for the purposes of 1.2.
The earlier reading of 1.2, before the February update, was narrower and caught social networks, forums, and comment threads. A handful of AI chat builders shipped through review in 2024 and 2025 without 1.2 safeguards because their product was framed as a one-on-one conversation with a model. App Review accepted some of those builds. The February 2026 clarification closes that gap. Patterns reported by developers on the Apple Developer Forums in March and April 2026 show that 1.2 rejection notes for AI chat apps have risen sharply.
What four safeguards does Guideline 1.2 actually require?
The text of Guideline 1.2 lists four required elements. The mapping to an AI chat app is what trips developers up.
| Safeguard | What it means in an AI chat app | What reviewers usually tap |
|---|---|---|
| Filtering | Blocks objectionable text from being posted, including user prompts and any shared model output | Submits a prompt with banned content and checks whether it lands |
| Reporting | Lets the user flag a specific message for human review | Long-press or three-dot menu on a model reply |
| Blocking | Prevents an abusive party (or persona) from continuing to message the user | Profile screen for another user, or settings for the AI persona |
| Contact information | A reachable support address visible inside the app | Settings, Help, or About screen |
The filtering bar is the one most often misread. The model's own safety filter from OpenAI or Anthropic sits upstream of the app and does not satisfy 1.2. Apple wants a filter inside the app that catches user-submitted content too, especially when prompts are stored, shared, or surfaced to other users. The OWASP MASVS input-validation requirements give a workable internal floor for what the filter should cover.
Where do Report and Block need to live in an AI chat UI?
Placement matters because reviewers test the same way real users tap. The expectation is straightforward:
- Report on every individual message, exposed through long-press or a three-dot menu next to the message bubble.
- Block on every counterparty surface: another user's profile in group chat, a group member list, or the settings screen for the AI persona.
A frequent mistake in AI chat apps is to place a single Report button at the conversation level. From the reviewer's screen that fails, because each reply is a separate piece of content. The actionable phrasing used in the nextnative App Store review checklist is to add a Report action to every content item and a Block user action to every user profile.
For AI chat apps where the only counterparty is a model, the Block action still belongs in the app. It can block the persona, mute the conversation for good, or end the session permanently. The point reviewers are looking for is that the user has a one-tap way to opt out of further contact with the offending entity.
For builders who want an outside check that the Report and Block surfaces map cleanly to the four safeguards before resubmitting, PTKD.com (https://ptkd.com) is one of the platforms focused on pre-submission scanning aligned with OWASP MASVS for AI-coded apps.
How does Apple expect reports to be handled?
The published guideline says "timely responses to concerns" and does not put a number on it. The Resolution Center cadence reported by developers from 2024 through 2026 sits around one business day, with twenty-four hours as the rule of thumb cited in third-party guides. Apple has not officially confirmed twenty-four hours, so treat it as a practical floor rather than a documented rule.
What that means for the build: the support address listed in App Store Connect under App Information needs to be monitored. A founder running an AI chat app from a personal email is fine; a generic alias that nobody reads will eventually catch a second 1.2 rejection when a follow-up report goes unanswered.
How do I answer a Guideline 1.2 rejection in the Resolution Center?
The rejection note in the Resolution Center usually quotes the guideline text and lists what the reviewer could not find. A common phrasing is missing required UGC precautions, no custom EULA, no content filtering, no user-blocking, no rapid response to reports.
Most 1.2 rejections can be answered with a single message in the Resolution Center pointing the reviewer to the screen where Report and Block live, with screenshots. A new build is needed only if a safeguard was genuinely absent from the previous binary.
The EULA is the one piece of the 1.2 surface that often catches AI chat apps. The Apple Standard EULA automatically prohibits objectionable content under section 9. If you attached a custom EULA in App Store Connect that does not include that clause, the rejection will point at the EULA before pointing at the buttons. The fix is metadata-only and happens in App Store Connect under App Information, with no new binary required.
What to watch out for
A few patterns trip up AI chat apps in particular.
- Streaming responses without a stable Report target. If the message bubble does not get a stable identifier until the model finishes streaming, the Report menu has to wait until the stream is complete. Some apps ship a Report action that crashes mid-stream. That is enough to trigger a rejection on its own.
- Group sessions where one user's prompt influences another user's view. This is UGC by any reading. Treat it that way and do not argue otherwise in the Resolution Center.
- Persona accounts with profile pictures. If the AI persona has a profile screen that mirrors a human profile, reviewers look for the Block button there. Match the social-app convention rather than fighting it.
- Treating the upstream model filter as the in-app filter. The model's content policy is one layer. Apple wants an in-app filter on top, especially for prompts that get stored or shared.
- A support address that nobody reads. The address in App Store Connect is part of the four safeguards. If a reviewer or a later complaint cannot get a reply through it, the same 1.2 enforcement will land again.
Key takeaways
- The February 2026 clarification means an AI chat app submitted without UGC safeguards is unlikely to pass review on the first attempt. Ship them before submitting, not after the rejection lands.
- Report belongs on every message; Block belongs on every counterparty surface, including the AI persona.
- The 24-hour response window is not officially published, but it matches the cadence developers see in the Resolution Center. Plan staffing for that, not for a weekly checkin.
- A short, screenshot-backed reply in the Resolution Center resolves most 1.2 rejections without a new build, provided the safeguards already exist in the binary.
- For teams that want an external read of their UGC implementation against the Guideline 1.2 checklist before submission, PTKD.com (https://ptkd.com) is one of the scanning platforms focused on pre-submission analysis aligned with OWASP MASVS for AI-coded apps.
