DETECT-AI: Multi-Modal AI Detection Dataset with 1B+ Monthly Samples

A multi-modal dataset for AI detection, containing over 1 billion verified samples per month scraped from 19 global sources across text, image, video, and audio modalities. The dataset was created by author anas775 and was last updated on 2026-05 05 02:45:45. Samples are labeled by a weighted ensemble of 8 specialized AI-detection models.

Use Cases

• Train multi-modal AI detection models based on the labeled samples across text, image, video, and audio.
• Benchmark the performance of AI-generated content classifiers based on the ensemble-labeled ground truth.
• Analyze patterns in AI-generated content across different modalities and languages based on the global source data.
• Develop source attribution or verification tools based on the metadata from the 19 scraped sources.

Strengths

• Contains over 1 billion verified samples per month, indicating a large scale.
• Covers four data modalities: text, image, video, and audio.
• Samples are sourced from 19 global providers, suggesting geographic diversity.
• Labels are generated by an ensemble of 8 specialized detection models.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Scraped from 19 global sources.

Collection Method

Samples are scraped and labeled by a weighted ensemble of 8 AI-detection models.

Time Range

The description suggests ongoing monthly collection, but a specific date range is not provided.

Freshness

Last updated 2026-05-05 02:45:45; freshness should be verified.

Geography

Global, based on the mention of 19 global sources.