To identify AI-generated or modified images from Google and third-party models, you can use the AI Content Detection API. This document explains how the API works, its limitations, and how to request access.
How it works
To use the API, send an image payload to the Gemini Enterprise Agent Platform endpoint using a REST API call. Supported image formats include the following:
- JPEG
- PNG
- WebP
Processing
The API processes the payload using machine learning models that analyze pixel-level artifacts, noise patterns, and spectral anomalies. The response aggregates findings from the models.
Data retention
The API does not store or retain processed images.
Request access
The AI Content Detection API is in Private Preview. To request access to the endpoint, complete the application form.
Disclaimers and limitations
The API provides a probabilistic estimate and doesn't guarantee definitive identification. Use of the service is subject to the following conditions:
- Statistical Errors: The model may produce false positives (authentic images flagged as AI-generated) and false negatives (AI-generated images misclassified as authentic).
- Advisory Use: The API output should serve as a supplementary signal. It must not be used as the sole basis for critical decisions, such as content takedowns or user sanctions.
- Liability: The service is provided "as is". Google assumes no liability for damages or disputes arising from reliance on API outputs.