Skip to content

AI companies

How does Apple source training data?

By SourceX Editorial · Updated

Short answer

Apple says Apple Intelligence models are trained on a mix of publicly available data crawled by its own Applebot web crawler, data licensed or purchased from third parties, open-sourced data, data from user studies, and synthetic data, with no private Apple user data included. Apple has also pursued multi-year licensing deals with publishers. This page reflects Apple's own disclosures and reported deals.

Apple's published training data page#

Apple's legal training data disclosure states that Apple trains generative AI models using publicly available data crawled by Applebot, data licensed or purchased from third parties, open-sourced data, data from user studies, and synthetic data, and that no private Apple user data is included.

Applebot and opt-outs#

Apple says Applebot does not crawl websites that require login or sit behind a paywall, and that web publishers can use robots.txt directives, including an Applebot-Extended tag, to prevent their content from being used to train Apple's generative AI models.

  1. 01Your work records
  2. 02Personal details removed
  3. 03You approve the release
  4. 04Licensed to AI developers
How it works with SourceX: Your work records → Personal details removed → You approve the release → Licensed to AI developers.Illustrative

Licensing deals with publishers#

The New York Times reported, as covered by TechCrunch, that Apple was in talks to license news archive content from publishers including Conde Nast, IAC and NBC News in multi-year deals reported to be worth at least fifty million dollars.

The Pile and YouTube subtitles controversy#

TechCrunch reported that Proof News found Apple had used a dataset called The Pile, containing YouTube video subtitles, to train some on-device models, and that Apple said it did not intend those specific models to power Apple Intelligence features.

What this means for a business#

A business publishing content online that does not want it used for Apple's AI training can apply the Applebot-Extended robots.txt directive, and should review Apple's published disclosure pages directly rather than assume a blanket opt-out covers every Apple AI feature.

SourceX and Apple#

SourceX is not affiliated with Apple. This page summarizes public reports; it does not mean Apple buys data through SourceX.

Frequently asked questions

Does Apple train on private user data?

Apple states no private Apple user data is included in training its generative AI models, according to its published training data disclosure.

Can websites opt out of Apple's AI training crawler?

Yes, Apple says publishers can use the Applebot-Extended directive in robots.txt to prevent their content from being used to train Apple's foundation models.

Has Apple paid publishers for training content?

Reports from the New York Times, covered by TechCrunch, describe Apple pursuing multi-year licensing deals worth at least fifty million dollars with several publishers.

Sources

Public sources, checked 2026-10-09. SourceX is not affiliated with the organizations named.

  1. Apple's stated training data sources: Applebot-crawled public data, licensed/purchased data, open-source, user studies, synthetic data, no private user data apple.com
  2. Applebot-Extended robots.txt opt-out for AI training support.apple.com
  3. Apple licensing deals with publishers and the Pile/YouTube subtitles controversy, reported 2024 techcrunch.com
  4. Apple multi-year licensing talks worth at least $50 million with publishers, reported 2023 appleinsider.com

Related resources

See if your company qualifies

A short company assessment. No data uploads are needed.

See if your company qualifies Talk to SourceX
See if your company qualifies