Veo could not generate videos because the input image violates … — Veo API support codes explained

Published 2026-09-23

Do not retry as-is

The exact error

Veo could not generate videos because the input image violates Agent Platform's usage guidelines. If you think this was an error, send feedback. Support codes: 15236754

Veo on Vertex rejected your request because a safety filter matched the input image, and gave you an eight-digit support code. Google publishes a table that maps each code to a safety category, which tells you what the filter objected to.

What it means

Google’s Responsible AI page for Veo on Gemini Enterprise Agent Platform (the new name for Vertex AI) says text, image and video inputs are assessed against safety filters, and that “the code listed in the output corresponds to a specific harmful category.”

The wording changed with the rename. The current page quotes the message with Agent Platform’s usage guidelines; earlier versions of the same page used Vertex AI’s usage guidelines. Both can appear in logs depending on when the request ran, so match on the start of the sentence and on the code, not on the product name.

The same page also covers a second, shorter input-side message: The prompt couldn’t be submitted or it might violate our policies. Google says that if the model responds with an error like this, “the input is triggering a safety filter”. It is the same mechanism without a code to look up.

Support code table

This is Google’s table, in the order it is published. The “What to check” column is our summary of the category description, not Google’s wording.

Support codes	Category	Google’s description	What to check
58061214, 17301594	Child	Rejects requests to generate content depicting children if `personGeneration` isn’t set to `"allow_all"` or if the project isn’t on the allowlist for this feature.	The `personGeneration` value and whether your project is allowlisted
29310472, 15236754	Celebrity	Rejects requests to generate a photorealistic representation of a prominent person or if the project isn’t on the allowlist for this feature.	Whether the image shows, or resembles, a prominent person; allowlist status
64151117, 42237218	Video safety violation	General safety violation.	The overall content of the image and prompt
62263041	Dangerous content	Potentially dangerous content.	Dangerous activities in the input
57734940, 22137204	Hate	Hate-related content.	Hateful content in the input
74803281, 29578790, 42876398	Other	Miscellaneous safety issues with the request	The whole request; no single category is named
89371032, 49114662, 63429089, 72817394, 60599140	Prohibited content	Prohibited content related to child safety or other sensitive content.	Content that cannot be generated at all
35561574, 35561575	Third-party content	Guardrails related to third-party content.	Material owned by others in the input
90789179, 43188360	Sexual	Sexual or suggestive content.	Sexual or suggestive content in the input
78610348	Toxic	Toxic content.	Toxic language in the prompt
61493863, 56562880	Violence	Violent content.	Violent content in the input
32635315	Vulgar	Vulgar content.	Vulgar content in the input

A code not in this table is undocumented; treat it as an unknown safety rejection rather than guessing.

How to confirm it’s this

The request went to Veo through Vertex (a Google Cloud project and region).
The message starts with “Veo could not generate videos because the input image violates” and ends with Support codes: and one or more numbers.
Look each number up in the table above. The example Google itself quotes, 15236754, is a Celebrity code.

The page is written for Veo on Vertex and does not say whether Gemini API filter messages carry these codes.

Fix

Find the category in the table: a setting, an allowlist, or the content itself.
Child codes: check personGeneration. Google’s description ties this category to that parameter and to an allowlist. Also check what your model and mode accept: the Gemini API Veo docs say image-to-video on Veo 3.1 accepts only "allow_adult", and that in EU, UK, CH and MENA locations allow_adult is the only allowed value.
Celebrity codes: check the image. See the celebrity and likeness page below.
Content categories: change the content. For the other categories the input itself is what the filter matched. Revise or replace it if it really falls in that category.
Report false positives. The message says to send feedback if you think it was an error. Include the support code; it is the reference Google’s page gives for the category.

This page will not help disguise content to get it past a filter. That would be working around a safety control.

Retry or not

No. Google describes this as the input triggering a safety filter, and nothing in the documentation points to a transient cause. Expect the same image and prompt to fail the same way.

For agents, log the support codes, map them to a category with the table, and surface that category: “rejected: Celebrity” is far more useful to the caller than a generic failure.

When it’s not this

videos were filtered out because they violated Google's Responsible AI practices. The generated output was blocked afterward; no support code.
The Gemini API celebrity message. “Sorry, we can’t create videos from input images containing celebrity or their likenesses.” It covers the same ground as the Celebrity codes but arrives as a filter reason.
Gemini image IMAGE_PROHIBITED_CONTENT. Code 42876398 is listed here as “Other” for Veo. The same number appears in a blocked Gemini image response quoted in litellm issue #20357; the Gemini image page covers that case.

Error text last verified against the sources below on 2026-09-23.