GPAI Ledger › Adobe Firefly (Adobe) › Capture 20 Aug 2026

Adobe Firefly — capture 20260820T105752Z

Verify: sha256sum 4bd528f863edf343f7780271463f34e72d9048ec5788a8645d2d044a58b29783.pdf must equal the hash above (the filename IS the expected hash); ots verify 4bd528f863edf343f7780271463f34e72d9048ec5788a8645d2d044a58b29783.pdf.ots -f 4bd528f863edf343f7780271463f34e72d9048ec5788a8645d2d044a58b29783.pdf (opentimestamps.org) proves the capture time (fresh proofs report 'pending' until bitcoin-anchored, typically within a day).

Extracted text

Machine-extracted text (layout may be lost; the authoritative content is the stored file above).

PUBLIC SUMMARY OF TRAINING CONTENT
Adobe Firefly
Image Model 5
November 2025

2
Public Summary of Training Content for General-
Purpose AI Models
for Adobe Firefly Image Model 5 as required by Article 53 (1)(d) of Regulation (EU) 2024/1689 (AI Act)
General Information
Provider identification
Provider name and
contact details
Adobe Inc.
345 Park Avenue, San Jose, CA 95110
Authorised representative name and
contact details
Adobe Systems Software Ireland Limited
Address: 4-6 Riverwalk, Citywest Business Campus, Saggart Dublin 24,
D24 DCW0, Dublin, Ireland
Email: euaiactauthorisedrep@adobe.com
Model identification
Versioned model name(s) Firefly Image Model 5
Model dependencies N/A
Date of placement of the model
on the Union market
October 28, 2025
Modalities, overall training data size and other characteristics
Modality Image and Video
Training data size • More than 1 billion images
• 10,000 to 1 million hours of video
Types of content • Image — Photography, illustrations, vector images, multi-layered images
• Video — Stock footage
Latest date of data: acquisition/collection
for model training
September 2025
Description of the linguistic characteristics
of the overall training data
The visual data include character and text elements from multiple languages.
Other relevant characteristics of the overall
training data
N/A
3
Additional comments (optional) N/A
List of Data Sources
Publicly available datasets
Have you used publicly available datasets
to train the model?
 Yes
If yes, specify the modality(ies) of the
content covered by the datasets concerned
 Image
List of large publicly available datasets OpenImages v7 https://storage.googleapis.com/openimages/web/index.html
OpenImages v7 is a large dataset of permissively licensed images with annotations.
General description of other publicly
available datasets not listed above
 Other datasets subject to permissive licenses.
Private non-publicly available datasets obtained from third parties
Datasets commercially licensed by rightsholders or their representatives
Have you concluded transactional
commercial licensing agreement(s) with
rightsholder(s) or with their representatives?
Yes
If yes, specify the modality(ies) of the
content covered by the datasets concerned
Image and video
Private datasets obtained from other third parties
Have you obtained private datasets
from third parties that are not licensed
as described in Section 2.2.1, such as
data obtained from providers of private
databases, or data intermediaries?
 Yes
If yes, specify the modality(ies) of the
content covered by the datasets concerned
Image and video
If publicly known, list private datasets
obtained from other third parties
N/A
4
Data crawled and scraped from online sources
Were crawlers used by the provider
or on behalf of?
No
Purposes of the crawler(s) N/A
General description of crawler behaviour N/A
Period of data collection March 2023 – May 2024
Comprehensive description of the type of
content and online sources crawled
Adobe did not crawl online sources. Instead, we searched specifically for content licensed under CC0
(license dedicating content to the public domain) or in the public domain in limited amounts within
select sites — such as photo hosting sites, educational websites, and museum websites — that
allowed for content downloading.
Type of modality covered Image
Summary of the most relevant domain
names crawled
Flickr.com
User data
Was data from user interactions with the AI
model (e.g. user input and prompts) used to
train the model?
No
Was data collected from user interactions
with the provider’s other services or
products used to train the model?
No
Synthetic data
Was synthetic AI-generated data created
by the provider or on their behalf to train
the model?
Yes
If yes, modality of the synthetic data Image
If yes, specify the general-purpose AI
model(s) used to generate the synthetic
data if available on the market
Firefly Image Model 5
Information about other AI models,
including provider’s own AI model(s) not
available on the market, used to generate
synthetic data to train the model to which
this Summary applies
Internal Adobe model
Other sources of data
Have data sources other than those
described in Sections 2.1 to 2.5 been
used to train the model?
No
© 2025 Adobe. All rights reserved.
Adobe and the Adobe logo are either registered trademarks or trademarks of Adobe
in the United States and/or other countries.
Data Processing Aspects
Respect of reservation of rights from text and data mining exception or limitation
Are you a Signatory to the Code of Practice
for general-purpose AI models that includes
commitments to respect reservations of
rights from the TDM exception or limitation?
No
Describe the measures implemented before
model training to respect reservations of
rights from the TDM exception or limitation
before and during data collection, including
the opt-out protocols and solutions
honoured by the provider or, as applicable,
by third parties from which datasets have
been obtained
N/A. TDM exception was not relied on for training data for this model.
Removal of illegal content
General description of measures taken Adobe has a policy to scan all visual content used for AI training with classifiers to identify and
remove CSAM. In addition, Adobe uses a combination of automated classifiers along with human
moderation designed to ensure that content used for training meets Adobe’s policies around harmful
content (see https://www.adobe.com/trust/transparency/content-policies/harmful-content.html  and
https://www.adobe.com/trust/transparency/content-policies/content-rights.html ).