---
title: "Introducing FILE type: a native column type for multimodal data"
description: "The beta of FILE introduces a native column type for storing unstructured data such as documents, images, audio, and video alongside structured table data. FILE applies fine-grained governance, row- and column-level controls, ABAC, and lifecycle synchronization, so deleting a row also deletes the corresponding binary in object storage. Its lightweight pointers fetch file content only when a query requires it, while SQL, Python UDFs, built-in AI functions, and materialized views support multimodal processing. The example uses sampled dashcam frames and object detection to identify stops with no hazard, then joins extracted results with trip metadata; the announcement also describes connectors, incremental pipelines, and an open effort around Parquet, Delta Lake, Apache Iceberg, and Apache Spark."
---

# Introducing FILE type: a native column type for multimodal data

[Databricks](https://yomu.fyi/company/databricks) · Michael Armbrust, Burak Yavuz, Dejan Krakovic, John Spencer · Aug 10, 2026

**Type:** Announcement

## Summary

The beta of FILE introduces a native column type for storing unstructured data such as documents, images, audio, and video alongside structured table data. FILE applies fine-grained governance, row- and column-level controls, ABAC, and lifecycle synchronization, so deleting a row also deletes the corresponding binary in object storage. Its lightweight pointers fetch file content only when a query requires it, while SQL, Python UDFs, built-in AI functions, and materialized views support multimodal processing. The example uses sampled dashcam frames and object detection to identify stops with no hazard, then joins extracted results with trip metadata; the announcement also describes connectors, incremental pipelines, and an open effort around Parquet, Delta Lake, Apache Iceberg, and Apache Spark.

## Context

Unstructured data such as contracts, images, recordings, and video is often managed separately from structured data, creating split governance and permission models, disconnected storage lifecycles, orphaned references, and compliance work. The announcement positions FILE as a way to make this data queryable, secured, and managed alongside structured data.

## Approach / What changed

FILE stores lightweight references to unstructured files as native table columns while retrieving content only when needed. It integrates with Unity Catalog access controls, SQL, Python UDFs, AI functions, materialized views, Spark Declarative Pipelines, and connectors; rows and their referenced binaries share a lifecycle, including deletion.

## Takeaways

- FILE columns use lightweight pointers rather than embedding large binaries in the table, so file content is fetched only for operations that require it.
- Unity Catalog secures FILE data with row-level, column-level, and attribute-based access controls, while deleting a row also deletes its file binary in object storage.
- The dashcam example samples video frames, applies object detection, and joins the resulting hazard data with trip metadata to find stops where no hazard was present.

**Tags:** [Data Pipelines](https://yomu.fyi/topic/data-pipelines), [Open Source](https://yomu.fyi/topic/open-source), [Spark](https://yomu.fyi/topic/spark)

- Source: [Databricks](https://www.databricks.com/blog/introducing-file-type-native-column-type-multimodal-data)
- Source URL: https://www.databricks.com/blog/introducing-file-type-native-column-type-multimodal-data
- Ingested by Yomu: 2026-08-30T16:51:52.254Z

[Read original post](https://www.databricks.com/blog/introducing-file-type-native-column-type-multimodal-data)