GPAI Ledger › Explanatory Notice + Template C(2025) 8311 final (European Commission) › Capture 11 Aug 2026

Explanatory Notice + Template C(2025) 8311 final — capture 20260811T110202Z

Verify: sha256sum 3245f6efbf00694a71bce67ea0b0c66a766428d1f631f9fcbfba9cb76f87dbc3.pdf must equal the hash above (the filename IS the expected hash); ots verify 3245f6efbf00694a71bce67ea0b0c66a766428d1f631f9fcbfba9cb76f87dbc3.pdf.ots -f 3245f6efbf00694a71bce67ea0b0c66a766428d1f631f9fcbfba9cb76f87dbc3.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).

EN   EN

EUROPEAN
COMMISSION
Brussels, 24.7.2025
C(2025) 5235 final
ANNEX

ANNEX

to the
Communication to the Commission
Approval of the content of the draft Communication from the Commission –
Explanatory Notice and Template for the Public Summary of Training Content for
general-purpose AI models required by Article 53 (1)(d) of Regulation (EU) 2024/1689
(AI Act)

1

1. Background

(1) Regulation (EU) 2024/1689 of the European Parliament and the Council of 13 June 2024 laying down
harmonised rules on artificial intelligence and amending certain regulations1 (‘the AI Act’) entered into
force on 1 August 2024. Section 2 of Chapter V AI Act lays down harmonised rules for providers of
general-purpose AI models , including obligations applicable to all providers of general -purpose AI
models2 and additional risk assessment and mitigation requirements for th ose of the most advanced
general-purpose AI models posing systemic risks3. Those rules will apply as of 2 August 2025.

(2) Article 53(1)(d) AI Act  requires all providers of general -purpose AI models to draw up and make
publicly available a  sufficiently detailed public summary of the content used for the training of the
model (the ‘Summary’), according to a template provided by the AI Office (the ‘Template’). Recital 107
AI Act  contains additional clarifications on the objectives of the Summary and the Template  which
include transparency on the data that is used for the training of general -purpose AI models, including
text and data protected by copyright law.

(3) Providers of all general-purpose AI models placed on the Union market must fulfil the above obligation,
including providers of general -purpose AI models released under free and open -source licenses4, in so
far as the model s fall within the  scope of the AI Act 5. Recital 107  AI Act  contains additional
clarifications on the Summary and the Template.

1 Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artif icial
intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 an d
(EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (Text with EEA relev ance),
PE/24/2024/REV/1, OJ L, 2024/1689, 12.7.2024.
2 Article 53 AI Act.
3 Article 55 AI Act.
4 The exception for free and open -source general-purpose AI model under Article 53(2) AI Act does not apply to the obligation to make
publicly available the Summary.
5 See Article 2 AI Act and Guidelines for providers of general-purpose AI models | Shaping Europe’s digital future

2

(4) Given that p roviders are obliged to draw up a Summary according to a Template provided by the
Commission, the latter holds important legal value for the proper implementation of the AI Act.  This
Explanatory Notice and the Template annexed to it aim to help providers of general-purpose AI models
comply with their obligation under Article 53(1)(d) of the AI Act  in a simple, consistent and effective
manner.

(5) The Template is based on the outcome of a multi -stakeholder consultation on general -purpose AI
models, organised by the AI Office from 30 July to 18 September 2024 6. Over 430 responses were
submitted from a wide range of stakeholders. Based on this input, the AI Office prepared and presented
its preliminary approach to the Template and allowed participants involved in the dra wing up of the
Code of Practice on General-Purpose AI7 to provide additional written feedback. The current version of
the Template annexed to th is Explanatory Notice  takes into account  comments received from 111
stakeholders, including providers of general-purpose AI models, business associations, rightsholders
organisations, academia, civil society and public authorities. The draft Template was also presented and
discussed with the AI Board Steering subgroup on General -Purpose AI and with the European
Parliament (IMCO-LIBE Committees) working group on AI.

2. Objective of the Summary

(6) General-purpose AI models are trained with large quantities of data for which there is typically limited
information available. Recital 107 AI Act explain s that the objective of the Summary is to increase
transparency on the content used for the training of general-purpose AI models, including text and data
protected by law  and to facilitate parties with legitimate interests , including rightsholders, to exercise
and enforce their rights under Union law.

6 AI Act: Have Your Say on Trustworthy General-Purpose AI | Shaping Europe’s digital future
7 General-Purpose AI Code of Practice | Shaping Europe’s digital future
RELEVANT LEGAL TEXT
Article 53(1)(d) AI Act.
Providers of general -purpose AI models shall […] draw up and make publicly available
a sufficiently detailed summary about the content used for training of the general -purpose AI
model, according to a template provided by the AI Office.
Recital 107 AI Act.
In order to increase transparency on the data that is used in the pre -training and training of
general-purpose AI models, including text and data protected by copyright law, it is adequate that
providers of such models draw up and make  publicly available a sufficiently detailed summary of
the content used for training the general -purpose AI model. While taking into due account the
need to protect trade secrets and confidential business information, this summary should be
generally compr ehensive in its scope instead of technically detailed to facilitate parties with
legitimate interests, including copyright holders, to exercise and enforce their rights under Union
law, for example by listing the main data collections or sets that went int o training the model,
such as large private or public databases or data archives, and by providing a narrative explanation
about other data sources used. It is appropriate for the AI Office to provide a  template for the
summary, which should be simple, eff ective, and allow the provider to provide the required
summary in narrative form.

3

(7) Such legitimate interests relate to copyright  and related rights and other intellectual property rights , but
also to other rights protected by Union law that should benefit from increased transparency.

(8) First, in relation to intellectual property rights, including copyright and related rights, transparency of
the data used for the model training should help rightsholders obtain relevant information on the content
used in the training of general -purpose AI models. This information is needed  to facilitate the exercise
of their fundamental right to intellectual property 8 and the fundamental right to an eff ective remedy in
the enforcement of their rights, as provided for in Union law in the area of intellectual property rights. In
the case of copyright and related rights, transparency of the training data will contribute to ensuring that
general-purpose AI models providers comply with Union law on copyright and related rights9.

(9) Second, transparency of the training data in the Summary may facilitate data subjects’ rights and more
broadly support the enforcement of the Union data protection rules . In particular, this can be done  by
summarising all the relevant information together, such as information  about the data scraped from the
internet or collected by the provider through interactions with the model or other services  and products.
The information  in the Summary  is not meant to replace, nor affect the respective information the
providers of general -purpose AI models  should make available to data subjects under Union data
protection law. In the context of the Summary , the interests of consumers and the protection of their
consumer rights under Union law may also be relevant.

(10) Third, transparency of the general characteristics of the content used for training may also assist
providers integrating these models into downstream applications to assess the diversity of the data. This,
in turn, will allow them to  implement, where appropriate,  mitigating measures  to ensure that the
fundamental rights to non-discrimination10 and language and cultural diversity11 are respected.

(11) Fourth, greater transparency of the training data may also facilitate the fundamental right to receive and
impart information12 and allow researchers to exercise their freedom of science 13 to conduct scientific
research. It can allow academic institutions and organisations to  critically evaluate the implications and
limitations of a particuler general-purpose AI model and the potential risks and harms associated with
the data used.

(12) Finally, transparency of the training data may also contribute to more transparent and competitive
markets. For example, information about whether publicly available general-purpose AI models have
been used to train other models, in particular through model distillation, or whether a model has been
trained on user data collected from provider’s own products and services, may help users and companies
better understand how their data and models have been used and avoid potential lock-in effects.

3. Comprehensive scope of the training data and sufficient details

(13) Information about the general-purpose AI model provided in the Summary should cover data used in all
stages of the model training, from pre -training to post -training, including model alignment and fine -
tuning. This covers all sources and types of data, regardless of whether the data are protected or not ,
including by an intellectual property right . Since Article 53(1)(d) AI Act mentions explicitly ‘training’,
other input data used during the model’s operation (e.g. through retrieval augmented generation) are not

8 Article 17(2) of the EU Charter of Fundamental Rights of the European Union, OJ C 326, 26.10.2012, p. 391–407.
9 Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital
Single Market and amending Directives 96/9/EC and 2001/29/EC (Text with EEA relevance.), PE/51/2019/REV/1, OJ L 130, 17.5.2019, p.
92–125.
10 Article 21 of the EU Charter of Fundamental Rights.
11 Article 22 of the EU Charter of Fundamental Rights.
12 Article 11(1) of the EU Charter of Fundamental Rights.
13 Article 13 of the EU Charter of Fundamental Rights.

4

required in the mandatory sections of the Template 14, unless the model actively learns from this input
data.

(14) Recital 107 AI Act explains that the information about the training content should be comprehensive in
its scope and sufficiently detailed to achieve the objective of the Summary of providing meaningful
public transparency and facilitat ing parties with legitimate interests to exercise and enforce their rights
under Union law.

(15) The Template annexed to th is Explanatory Notice aims to provide a common minimal baseline for the
information to be made publicly available in the Summary. It consists of three main sections:

1. General information: this section requires information allowing identification of the provider
and of the model, and information on modalities, the size of each  modality within broad
ranges, as well as general characteristics of the training data.
2. List of data sources:  this section requires disclosure of the main datasets that were used to
train the model, such as large private or public databases, and a  comprehensive narrative
description of the data scraped online by or on behalf of the provider (including a summary of
the most relevant domain names scraped) and a narrative description of all other data sources
used (e.g. user data or synthetic data) to ensure completeness of the summary regarding the
content used for the model training15.
3. Relevant data processing aspects: this section of the Template requires disclosure of certain
data processing aspects that are relevant for the exercise of the rights of parties with legitimate
interests under Union law . This is especially important for compliance with Union law  on
copyright and related rights and for the removal of illegal content to mitigate the risk that such
illegal content may be reproduced and disseminated at scale by the general-purpose AI model.

(16) On the basis that the Summary  aims to provide sufficient details and to facilitate parties with legitimate
interests, including rightsholders, exercising their rights under Union law, the Template requires a
disclosure of a summary of the list of top domain names crawled and scraped from online sources in a
summarised narrative form 16. At the same time, it does not require disclosure of the details for the
specific data and works used to train the model as this would go beyond the requirement in Articl e
53(1)(d) to provide just a ‘summary’ , which in line with Recital 107 AI Act must be “generally
comprehensive” but not “technically detailed”. Providers may nevertheless decide on a voluntary basis
to go beyond the minimum requirements in the Template and disclose in the Summary more details than
what is required by Article 53(1)(d) AI Act and the Template17. Furthermore, for domain names scraped
or crawled from the inte rnet that are not listed in the  Summary, it is recommended that providers act in
good faith and on a voluntary basis enable parties with a legitimate interest including rightholders, upon
requests, to obtain information whether the provider has scraped and used for training content which
includes protected works and other subject matter  that rightholders have made available on specific
internet domains . This recommended and voluntary ‘upon request’ mechanism does not affect other
available remedies for rightholders under Union law on enforcement of intellectual property rights (e.g.
Article 8 of the Intellectual Property Rights Enforcement Directive)18.

14 Since such data is used as input for the generation process, its influence on the outputs of the model may be significant, and relevant for the
exercise of the rights of parties with legitimate interes ts. For this reason it may be disclosed by the provider on a voluntary basis in the
optional Sections in the Template which allow the provision of additional information.
15 See in this context Recital 107 of the AI Act.
16 See Section 2.3. of the Template that requires a list of the internet domain names (top and second-level domain, e.g. “example.com”) in the
top 10 % of all domain names determined by the size of the content scraped (in a representative manner across modalities, where applicable).
For small and medium-sized enterprises (SMEs), including start-ups, the Template requires the internet domain names in the top 5%, or the
top 1000 domains to ensure proportionality of the burden on SMEs in line with recital 109 AI Act.
17 See in the end of each sub-section of the Template, optional part with possibilities to provide additional information.
18 Directive 2004/48/EC of the European Parliament and of the Council of 29 April 2004 on the enforcement of intellectual property rights
(OJ L 157, 30.4.2004), OJ L 195, 2.6.2004, p. 16–25.

5

4. Balance with trade secrets and confidential business information

(17) As explained in Recital 107 AI Act, the Template should seek to strike a balance between serving the
interests of parties with legitimate interests and promoting increased transparency of the training content
in a meaningful way, while respecting the rights of all parties concerned, in particular taking due
account of the need to protect trade secrets and confidential business information.

(18) Since the Commission is bound by the Charter on fundamental rights, this careful balancing exercise has
been implemented in relation to the information that the Template requires to be disclosed in order for
providers to fulfil their obligation under Article 53(1)(d) AI Act and provide a ‘sufficiently detailed’
public summary of the training content. The provision of information regarding more specific details
about the content used for the training of the general -purpose AI models is required in the Templa te
only where it is necessary to enable the exercise of rights protected under Union law in a meaningful
manner as required by Article 53(1)(d) and Recital 107 AI Act. Determining which details should be
disclosed has been the result of a careful balancing  exercise carried out by the Commission in drawing
up the Template to ensure that relevant information on the training data is provided to meet the
Template’s objectives, while confidential commercially sensitive information about the data sources and
the precise manner in which providers curate the data and train their models is preserved.

(19) To protect providers’ trade secrets, different levels o f detail are required in the Template depending on
the source of data considered. In particular, limited disclosure of information is required for licensed
data given that the rightsholders concerned are parties to the licensing agreements (see Section 2.2.1 of
the Template). Furthermore, private datasets not commercially licen sed by rightsholders and obtained
from other third parties have to be listed only if publicly known (or the provider wants to make them
publicly known), and otherwise described in a g eneral manner (see Section 2.2.2 of the Template).
Considering the public nature of the information contained in publicly available datasets, more detail is
required about those datasets, including the disclosure of ‘large’ datasets (defined in th e Template), in
line with Recital 107 AI Act (see Section 2.1 of the Template).

(20) For data scraped from online sources, the Template requires disclosure of relevant information such as
the crawlers used, their purpose and behaviour, the period of collection and a comprehensive description
of the type of content and online sources scraped (see Section 2.3 of the Template). In addition, the
Template requires disclosure of a summary list of  most relevant  domain names crawled and scraped
from online sources by or on behalf of the provider in a summarized narrative form19, in so far as their
content has been used for the training of the general -purpose AI model. Such a summary of the domain
names scraped from the internet aims to provide a meaningful information about the most relevant top
domain names scraped, while striking a balance with  the trade secrets ensuring the Summary remains
non-technical and in a summarised narrative form, as required by Recital 107 AI Act.

(21) Furthermore, the Template (see Section 2.4) requires minimal information about user data collected
through user interactions with all services and products of the provider, including interactions with the
providers’ AI models. This category excludes data licensed by users based on commercial  transactional
agreements already covered under Section 2.2.1, or customer data used for fine -tuning a model for
specific purposes. For synthetic data generated by AI model(s) used for training purposes and in
particular for model distillation (see Section 2.5 of the Template), the information is also limited to
names of the general-purpose AI model(s) used if th ose models have been placed on the market or , if
other AI models have been used, including models owned by the providers, information about the model
(including a general description of the model training data if known and in so far as this may be needed
for the exercise of the rights of parties with legitimate inetersts and to avoid circumvention of the
disclosure obligations in the other Sections of the Template).

19 See footnote 16 above and Section 2.3. of the Template.

6

(22) The Template does not require disclosure of the exact mix and composition of data sources , but only
high-level information about the training data size per modality (selection amongst three very broad
ranges) and aggregated across all sources20 (see Section 1.2 of the Template).

5. Simple, uniform and effective reporting

(23) The information requested by the Template is to be provided in a narrative, simple and effective form .
The Template aims to ensure the reported information is useful and understandable to the public and to
the parties concerned, while avoiding unnecessary bur den on providers of general -purpose AI models ,
including SMEs.

(24) Each Section of the Template includes clear and short instructions to allow providers to report the
required information in an easy and uniform manner. The Commission aims to provide the Template as
an online form and to publish it on its website.

(25) Providers should ensure that the information included in the Summary is reported in good faith  and in
an accurate and comprehensive manner. Flexibility is provide d under specific sections, as indicated in
the Template, to disclose only information that is relevant, necessary for the purpose of the Summary,
and practicable to obtain (e.g. regarding the categorisation of some of the content or the characteristics
of the training data, or the period of data collection).

(26) The AI Office may verify whether the Template has been filled in correctly  in order to assess if the
provider has complied with Article 53(1)(d) AI Act . In this context, the AI Office  has all enforcement
powers under the AI Act and and may request corrective measures. Non-compliance may be sanctioned
with fines of up to 3% of the provider’s annual total worldwide turnover in the preceding financial year
or EUR 15 000 000, whichever is higher. The lawfu l collection and processing of the data remains the
responsibility of the provider under other applicable Union law (e.g. copyright and data protection). The
AI Office will supervise the implementation of the obligation to provide a compliant summary under
Article 53(1)(d) AI Act without performing a work-by-work assessment or checks whether specific
content has been used or not for the training of the general-purpose AI model (Recital 108 AI Act).

(27) In case of disputes, providers and parties with a legitimate interest , including rightsholders, are
encouraged to use alternative dispute resolution mechanisms available at national level (such as
mediation) and other available remedies provided for by Union and national law (e.g. under Article 8 o f
the Intellectual Property Rights Enforcement Directive)21.

6. Modifications of existing general-purpose AI models and updates

(28) An existing general -purpose AI model already placed on the Union market may be  modified by a
downstream entity in such a way that the downstream entity becomes the provider of the resulting
general-purpose AI model, as specified in the Commission guidelines on General-Purpose AI models22.
In such cases, the information reported by the modifying entity in the Template should be limited to the
training content used for the model modification only23, and the name of the model(s) that was modified
should be clearly indicated in the Summary (see Section 1.2 of the Template).

(29) The Summary should also be updated whenever  the provider further train s its own general-purpose AI
model placed on the market on additional data that requires an update of the content of the Summary. In
those cases, the Summary should be updated at six-month intervanls or if in the meantime the additional

20 Ref to ECJ case-law aligned with this approach.
21 Directive 2004/48/EC of the European Parliament and of the Council of 29 April 2004 on the enforcement of intellectual property rights
(OJ L 157, 30.4.2004), OJ L 195, 2.6.2004, p. 16–25.
22 Guidelines for providers of general-purpose AI models | Shaping Europe’s digital future
23 See also Recital 107 AI Act.

7

data used to further train the model  requires a material ly significant  update of the content of the
Summary, whichever is sooner. In such cases, the Summary should be updated to reflect this additional
data, as well as the date of the update. The updated Summary should be made publicly available
alongside the modified model.

(30) The same Summary may be used for diff erent models or different model versions if the content of their
respective Summaries is identical. In this case, the Summary should clearly specify the different models
and model version to which it applies. In addition, if different models or model versi ons are based on
the same general -purpose AI model that has already been placed on the Union market, and the
Summaries for each model and model version are different so that they cannot be covered by a single
Summary, the Summaries for each of those models  or model versions only need to cover the training
data specifically used to further modify (including fine-tune) them out of the original model. In this
case, a clear reference should be made to the original model in the Summary for each relevant model or
model version, and a link to the Summary of the original model included (see Section 1.2).

(31) Where the same Summary is used for multiple models or model versions, in accordance with point (30),
reference in the template to the ‘model’ should be understood as a reference to each model or model
version covered by the Summary. Reference to ‘training data’ should be understood as a reference to the
training data for each of these models or model versions.

7. Publication of the Summary

(32) The Summary should be made publicly available at the latest when the model is placed on the Union
market. It should be published on the provider’s official website in a clearly visible and accessible
manner, clearly indicating which model(s) (and possibly model version(s)) the Summary covers subject
to the conditions specified in point (30) above. The Summary should also be made publicly available
together with the model across all its public distribution channels (e.g. online platforms).

8. Entry into application of the obligation and special rules for models
placed on the market before 2 August 2025

(33) The obligation for making the Summaries publicly available becomes applicable as of 2 August 2025.
For models placed on the market before 2 August 2025, providers should take the necessary steps to
make the corresponding Summary publicly available no later than 2 August 2027. Where a provider of
a model placed on the market before 2 August 2025  cannot, despite their best efforts, provide parts of
the information required to prepare the Summary because the information is not available or its retrieval
would impose a disproportionate burden on the provider, the provider should clearly state and justify the
corresponding information gaps in its Summary24. The supervision and enforcement by the AI Office for
compliance with the rules for general-purpose AI models will start as of 2 August 2026.

9. Review of the Explanatory Notice and the Template

(34) The Commission will monitor the implementation of the Template  annexed to this Explanatory Notice
and where necessary review the Notice and the Template, in view of practical experience gained and of
the pace of technological, societal and market developments in this area.  If the Commission deems it
necessary, such a review may take place before the entry into application of the enforcement powers of
the AI Office on 2 August 2026.

24 See under each Section of the Template a box for possible additional information (optional).

8

Annex
Template for the Public Summary of Training Content for
General-Purpose AI models required by Article 53 (1)(d)
of Regulation (EU) 2024/1689 (AI Act)
Version of the Summary:

Version of the summary, with link(s) to previous versions where applicable
Last update:  DD/MM/YY
1. General information
1.1. Provider identification
Provider name and
contact details:
Authorised representative
name and contact details:
Only applicable if the provider is established outside the Union ( see Article
54 AI Act).

1.2. Model identification
Versioned model name(s):
Provide the unique identifier(s) for the model (s) or model version(s) covered
by this Summary  (e.g. Llama 3.1 -405B). In accordance with point 30 of the
Commission Explanatory Notice to the Template , the same Summary may be
used for different model(s) or model version(s)  provided the content of the ir
respective Summaries is identical. Where available, provide link(s) to
additional publicly availabl e documentation, such as the model card, for the
model(s) or model version(s).
Model dependencies:
If the model is the result of a modification , including  fine-tuning, of one or
more general-purpose AI models already placed on the Union market, specify
the model (version) name(s) of that/those models and provide a link to their
Summary(ies) where available.
Date of placement of the
model on the Union
market:
Indicate t he date  on which the model was placed on the U nion market
(including the dates each model (version(s)) was placed on the market , if the
Summary applies to more than one model  or version (see point 30 of the
Commission Explanatory Notice to the Template).

1.3. Modalities, overall training data size and other characteristics
This section requires general information about the overall training data after pre-processing and before the
training of the model.

Modality
Select the
modalities present
in the training data,
to the extent that
they are identifiable
Training data size
For each selected modality, select the range within
which the estimated total training data size for
that modality falls. Dynamic datasets may be
excluded from the estimation.
Types of content
For each selected modality, provide a general
description of the type of content that has been
included in the training data.

☐ Text
☐  Less than 1 billion tokens
☐ 1billion to10 trillions tokens
☐ More than 10 trillions tokens

Alternatively, specify the approximate size
in a different measurement unit:
_____________

Examples of possible types of content include
fiction and non fiction text, scientific text, press
publications, legal and official documents, social
media comments, source code.
☐ Image
☐ Less than 1 million images
☐ 1Million to1 billion images
Examples of possible types of content include
photography, visual art works, infographics,
social media images, logos, brands.

9

☐ More than 1 billion images
☐ Audio25
☐ Less than 10 000 hours
☐ 10 000 to1 million hours
☐ More than 1 million hours
Examples of possible types of content include
musical compositions and recordings, audiobooks,
radio shows and podcasts, private audio
communication.

☐ Video
☐ Less than 10 000 hours
☐ 10 000 to1 million hours
☐ More than 1 million hours
Examples of possible types of content include
music videos, films, TV programmes,
performances, video games, video clips,
journalistic videos, social media videos.
☐ Other
Specify the modality and for each one
indicate approximate size and unit of
measurement

Latest date of data:
acquisition/collection for
model training:
Indicate the latest date when data was collected/obtained for the model
training:  MM/YYYY
Additionally, i ndicate if the model is continuously trained on new  or
dynamic data after this date.

Description of the linguistic
characteristics of the overall
training data:

Where applicable , describe the languages covered by the training data
(e.g., text, videos or speech),  focusing in particular on EU official
languages.
Other relevant characteristics
of the overall training data:
Where such information is readily available and in so far as it is relevant
and practicable, describe other relevant characteristics of the overall
training data, such as national/regional or demographic specificities of
the training data.
Additional comments
(optional):
Providers may also disclose other relevant information on a voluntary
basis (e.g. the compression or tokenization methodologies applied for the
data size calculation, the samp ling frequency/rate plays for audio or
video content).

2. List of data sources
This section requires information about specific sources of data used to train the general -purpose AI model. In
this section “dataset” should be understood as a single, pre -packaged collection of data. The filtering and pre -
processing of data collected from the same pre-packaged collection should not be considered a new dataset to be
disclosed separately in the sections below . If a particular dataset can be assigned to more than one of the
categories below, providers should select the most relevant category and only report the dataset in that
category, except in the case of synthetic data (see Section 2.5).

2.1. Publicly available datasets
This section requires information about datasets that were used to  train the model and which have been
compiled by a third party, are made available publicly for free, and are readily downloadable as a whole or in
predefined chunks, such as datasets and collections available on public repositories and online platforms,
specialised websites, or snapshots of common crawl. The public availability of the datasets for free does not
mean that the content at issue is necessarily free of rights since it may be subject to licensing arrangements or
conditions of use (e.g., certain f ree and/or open licenses may determine the scope of the uses, including
prohibiting uses relating to model training).

A dataset is considered to be “large” if the total data size for any one of the modalities contained in the dataset
exceeds 3% of the siz e of all publicly available datasets for that modality used for training. The size of the
dataset should be based on its size after pre -processing (for example filtering), and without splitting the dataset
to prevent reporting circumvention.
Have you used publicly available datasets totrain the model?   ☐ Yes     ☐ No

25 Excluding audio that is part of video, as this should be reported under the “video” modality instead. Furthermore, the Commission
understands the modality of ‘audio’ to include ‘speech’.

10

If yes, specify the modality(ies) of
the content covered by the datasets
concerned:
☐ Text   ☐ Image   ☐ Video  ☐ Audio ☐ Other (please
specify)
List of large publicly available
datasets:
For each large dataset, provide the identifier/name of the dataset
and a link through which the dataset can be accessed . If a link is not
available, provide a general description of the dataset, including the
approximate start and end dates of the data collection if known
(otherwise indicate “not known” ). If only part of the datasets ha s
been used for the training, indicate the general approach to selecting
those parts.
General description of other
publicly available datasets not
listed above:
For other publicly available datasets that are not listed above,
provide a general description of their content . The description could
include indication of: (i) the types of modality (e.g. text, images) , (ii)
nature of the content (e.g. personal data, copyright protected
content, machine generated data such as Internet of Things or
synthetic data), (iii) its linguistic characteristics, where applciablem
(iv) the approximate start and end dates of the data coll ection if
known (otherwise indicate “not known”).
Additional comments (optional):
Providers may also disclose other relevant information  on a
voluntary basis, e.g. size of the datasets and other relevant details.

2.2. Private non-publicly available datasets obtained from third parties
This section requires information about private non-publicly available datasets of third parties that are not
publicly available and not disclosed under Section 2.1. These include:
1) datasets for which transactional commercial licensing agreements were concluded between the
provider and the rightsholders or their representatives, including by collective management
organisations and legitimate content aggregators who have the right to collective ly license works on
behalf of rightsholders (Section 2.2.1);
2) other private datasets obtained through data intermediaries, non -publicly available databases and
datasets of third parties for which transactional commercial licenses have not been concluded with
rightsholders or their representatives (Section 2.2.2).

2.2.1. Datasets commercially licensed by rightsholders or their representatives
Have you concluded transactional commercial licensing
agreement(s) with rightsholder(s) or with t heir
representatives? ☐ Yes     ☐ No
If yes, specify the modality(ies) of the content covered
by the datasets concerned:
☐ Text   ☐ Image   ☐ Video
☐ Audio ☐ Other (please specify)

2.2.2. 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     ☐ No
If yes, specify the modality(ies) of the
content covered by the datasets concerned:
☐ Text   ☐ Image   ☐ Video  ☐ Audio ☐ Other (please
specify)
If publicly known, list private datasets
obtained from other third parties:
If publicly known, list the identifiers/names of the main private
datasets from third parties that are not licensed as described in
Section 2.2.1 and that are used to train the model, and provide
links to relevant information, where available.

11

General description of non -publicly known
private datasets obtained from third parties
For those private datasets used to train the model that are not
publicly known and whose identifiers are not listed ab ove,
provide a general description of their content . The description
should indicate (i) the modalities (e.g,. text, images), (ii) nature
of the content (e.g., personal data, copyright protected content,
machine generated data such as Internet of Things or synthetic
data) and (iii) its linguistic characteristics, where applicable.
Additional comments (optional):
Providers may also disclose other relevant information  on a
voluntary basis , e.g. the period of data collection, size of the
datasets and further details.

2.3. Data crawled and scraped from online sources
This section requires information about crawled, scraped data, orotherwise compiled from online sources
directly by the provider of the model or on their behalf (i.e. excluding publicly available datasets already
compiled by third parties and made available on platforms such as  common crawl that are covered under
Section 2.1).
Were crawlers used by the provider or
on behalf of? ☐ Yes     ☐ No
If yes, specify crawler
name(s)/identifier(s):
Purposes of the crawler(s):
General description of crawler
behaviour:
For example, this includes respect of captchas, password
protected websites and paywalls, respect of robot.txt and other
protocols, while crawling.
Period of data collection: From MM/YYYY to MM/YYYY
Comprehensive description of the type
of content and online sources crawled:
Provide a comprehensive description of the type of content
crawled, including its geographical, linguistic or demographic
characteristics, as well as an indication of the type of websites
scraped (e.g. news, blogs, social media, forums, community
websites, other user -generated content platforms, websites  of
cultural heritage institutions , educational sites, government
portals, personal blogs, streaming, ga ming platforms, online
TV platforms, synthetic data libraries, etc).
Type of modality covered:
 ☐ Text   ☐ Image   ☐ Video  ☐ Audio ☐ Other (please
specify)
Summary of the most relevant domain
names crawled:
In so far as any content from internet domains has been
crawled or scraped and used for the training of the model,
provide a list of those most relevant internet domains names
(top and second -level domain, e.g. “example.com”) by listing
the top 10 % of all  domain names determined by the size of the
content scraped (in a representative manner across all
modalities where applicable) . S mall and medium -sized
enterprises (SMEs), including start-ups, should disclose top 5%
of all domain names or 1 000 internet domain names,
whichever is lower . You can provide t his list for instance as a
downloadable file or provide here information on how to
access it.
Additional comments (optional):
Providers may also disclose other relevant information  on a
voluntary basis , for instance more domain names than those
required in the list above and/or URLs and the sources of
individual works.
2.4. User data
This section requires information about user data collected by all services and products of the provider,
including through mail services, social media platforms, content platforms or interaction with the providers’ AI
models and/or systems.  This does not cover data licensed by users based on commercial transactional
agreements described inSection 2.2.1., or customer data to fine-tune models for specific purposes.

12

Was data from user interactions with the AI model
(e.g. user input and prompts) used to train the
model?

☐ Yes     ☐ No
Was data collected from user interactions with the
orovider’s other services or products used to train
the model?

☐ Yes     ☐ No
If yes, provide a general description of the
provider’s services  or products that were used to
collect the user data:

Type of modality covered:
☐ Text   ☐ Image   ☐ Video  ☐ Audio
☐ Other (please specify)
Additional comments (optional): Providers may also disclose other relevant
information on a voluntary basis.

2.5. Synthetic data
This Section requires information about synthetic data created by or on behalf of the provider for training the
model directly on the outputs of another AI model, in particular through model distillation or model alignment
(e.g. AI feedback through reinforcement learni ng). This does not include the use of AI models to clean or enrich
data (e.g. AI-generated metadata to enrich or modify a dataset, such as creating depth maps or text descriptions
of image s). In case  this concerns publicly available datasets as described in Section 2.1, these should be
reported in that Section of the Template. In case this concerns synthetic datasets created by third parties on
behalf of the provider, these should be reported in this Section of the Template instead of in Section 2.2.2.

Was synthetic AI -generated data created by the
provider or on their behalf to train the model?

☐ Yes     ☐ No
If yes, modality of the synthetic data:  ☐ Text   ☐ Image   ☐ Video  ☐ Audio ☐
Other (please specify)

If yes, specify the general-purpose AI model(s)
used to generate the synthetic data if available on
the market:

Specify the name of the general -purpose AI
model(s) and provide a link to their Summary(ies)
where available.

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:
Provide information about other AI models used to
generate synthetic data, including provider’s own
models if not available on the market . This
includes a general description of the model
training data if known and in so far as this may be
needed for the exercise of the rights of parties with
legitimate inetersts and to avoid circumvention of
the disclosure obligations in the other Sections of
the Template.
Additional comments (optional): Providers may also disclose on a voluntary basis
other relevant information.

2.6. Other sources of data
This Section requires information about  data that does not fall under any of the categories in the previous
Sections, for example data collected from offline sources, self -digitised media (e.g., digitised analog text context,
images), datasets labelled by humans  commissioned by the provider , or human gene rated data through
reinforcement learning.

Have data sources other than those described in
Sections 2.1 to 2.5 been used to train the model?
☐ Yes     ☐ No

13

If yes, provide a narrative description of these data
sources and the data:

Additional comments (optional): Providers may also disclose other relevant
information on a voluntary basis.

 3. Data processing aspects
3.1. Respect of reservation of rights from text and data mining exception or
limitation
This Section concerns measures implemented by the provider to identify and comply with the reservation of
rights from the text and data mining (TDM) exception or limitation expressed pursuant to Article 4(3) of
Directive (EU) 2019/790, as outlined in the copyrig ht policy put in place by the provider in accordance with
Article 53(1)(c) AI Act.

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?

☐ Yes     ☐ 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 hono ured by the
provider or , as applicable,  by third parties from which
datasets have been obtained:

Additional comments (optional):
Providers may also disclose other relevant
information on a voluntary basis. Providers
are also encouraged to disclose a summary
of their copyright policy under Article
53(1)(c) AI Act, if made pub licly available
(e.g., by providing a link to the relevant
web-page).

3.2. Removal of illegal content
This Section concerns measures taken to avoid or remove illegal content under Union law from the training data
(such as blacklists, keywords, and model-based classifiers), without requiring disclosure of specific details about
the provider’s internal business practices or trade secrets . Such measures are advisable if the training data is
likely to include illegal or unlawful content under Union law, in particular child sexual abuse material and
terrorist content and the non-authorised use of material protected by intellectual property rights. Such measures
do not include data selection practices, for example to increase the capability of the model.

General description of measures taken:

3.3. Other information (optional)

Other relevant information about data
processing (optional):

Providers are also encouraged to disclose on a voluntary basis
other relevant information about relevant data processing
aspects and measures taken before or after the training of the
model that is relevant for the respect and exercise of rights
protected under Union law.