AI Supplementary Agreement
on AI-supported Data Processing on Crazypatterns (AISYS, SIA)
As of March 5, 2026
Role clause and validity
For the sake of linguistic simplicity and better readability, terms such as "user" or "author" are predominantly used in the masculine form in this supplementary agreement. These terms are, of course, intended to apply equally to all genders (m/f/d) and do not imply any discrimination.
This supplementary agreement consists of Module A (General Provisions) and Module B (Additional Provisions for Authors). Module A applies to all users of the Crazypatterns platform. Module B also applies to users who upgrade to an author account; the provisions of Module B are displayed separately during the upgrade and are expressly accepted by the corresponding confirmation action.
Module A – General Provisions
1. Principles, Subject Matter, Inclusion, and Priority
1.1 Principles ("Privacy First"). The AI functions on Crazypatterns are developed and operated in such a way that they improve the user experience, assist in finding suitable content, and serve the security and quality of the platform. The focus is on data protection. Users retain control over activated AI functions within the scope of the available settings. Data processing by external providers is carried out exclusively on our behalf as the Controller; use for the improvement, training, or fine-tuning of models generally provided by the provider (basic models) is excluded.
1.2 Subject matter. This supplementary agreement additionally regulates AI-supported data processing and other automated evaluations for additional functions of the platform.
1.3 Inclusion. This supplementary agreement applies in addition to the User Terms and Conditions; and, for authors, additionally to the Author Terms and Conditions.
1.4 Priority. Insofar as this supplementary agreement contains deviating or more specific provisions, these shall take precedence over the User Terms and Conditions/Author Terms and Conditions for the matters regulated herein.
2. Terms used in this AI supplementary agreement
2.1 AI systems are machine-based systems that process data with a certain degree of autonomy and generate results such as predictions, recommendations, content, or decisions.
2.2 AI evaluations are automated analyses of content to generate or derive information, including, but not limited to, inference and analysis, creation of vector or feature representations, and training and fine-tuning of models, each including functionally similar further developments or successor technologies, regardless of whether ML or non-ML is used.
2.3 Internal AI systems are systems operated by AISYS in server or data center environments without processing by external providers.
2.4 External providers are service providers bound by instructions who are used to perform additional functions—usually cloud AI services.
2.5 Product presentation data is content about the product posted by the author, separated into
- (a) text content (e.g., description, title, tags, category, file types, price) and
- (b) images and preview videos (not full product videos).
They do not include the digital products themselves (see 2.6).
2.6 Digital products are the files provided by the author that are available for sale, either for a fee or free of charge (e.g., instructions, templates, PDFs, complete product videos, or ZIP archives with product files).
2.7 Summaries are highly simplified, non-reproducible summaries or descriptions of the features of digital products or product presentation data that contain only general or functional information (e.g., technology, material, difficulty level, subject, or category) and do not allow for the reconstruction or restoration of the complete content.
2.8 Opt-out refers to the deactivation of certain AI processing declared by the user or author via the respective user account or author account.
2.9 PII (personally identifiable information) is any information relating to an identified or identifiable natural person (e.g., name, email address, face).
2.10 User-generated content (UGC) refers to all content, data, and interactions posted by the user on the platform and at least partially visible to other users (e.g., forum posts, comments, Q&A, reviews, uploaded images, "Like"/"Helpful" marks).
2.11 Private user data is content and data created by the user that, by its nature, is not intended for public areas of the platform or for other users (e.g., private messages, product watch lists, shopping cart contents).
2.12 Additional functions are all automated or AI-supported processing and services that go beyond the basic functions of the platform and serve to improve the use, quality, security, or functionality of Crazypatterns. These include, in particular, but are not limited to, functions for content analysis, quality assurance and evaluation (including scoring), moderation, search, sorting, personalization, translation, recommendation, categorization, abuse and fraud prevention, and the development and optimization of new platform functions.
2.13 Anonymization refers to the processing of personal data in such a way that the data subject can no longer be identified; Re-identification is not possible by us or by third parties using legally permissible means.
2.14 Pseudonymization refers to the processing of personal data in such a way that it can no longer be attributed to a specific person without the use of additional information, provided that this additional information is kept separately and is subject to technical and organizational measures that prevent re-identification.
2.15 PII gate refers to a technical system for the automated detection of personal data (PII) in content before it is further processed as part of the additional functions. Detection is performed automatically using technical patterns and may be incomplete or incorrect in individual cases.
2.16 Inference refers to the ongoing processing of content by AI systems to generate results such as suggestions, ratings, or classifications without retraining the underlying model.
2.17 Vectorization and indexing, including so-called "embeddings," refer to the conversion of content into feature or vector representations to make it findable and comparable for search, sorting, recommendations, or similar additional functions.
2.18 Internal training refers to the adaptation or improvement of internal AI models (e.g., training, fine-tuning).
2.19 External training refers to the training or fine-tuning of external models.
2.20 Usage data is log and interaction data generated by the use of the platform, separate from UGC (2.10) and private user data (2.11). This includes, in particular, pages viewed and search queries, click and scroll events, interactions (e.g., likes, watch lists, downloads), as well as purchase/order/cancellation and shopping cart transactions, each with a timestamp and, if applicable, pseudonymous account reference (e.g., hashed IDs).
2.21 AI-generated or AI-modified media are images or videos that have been created in whole or in part by AI systems or in which essential image elements—including backgrounds—have been replaced, added, or significantly altered with the aid of AI; the term includes media within UGC (2.10), product presentation data (2.5), and media components of digital products (2.6).
3. Granting of rights for additional functions and obligations of the user
3.1 In addition to the rights granted in Sections 12.3–12.6 of the User Terms and Conditions, you grant us the right to use your user-generated content (UGC) within the scope of the additional functions, in particular for automated analyses (AI evaluations).
3.2 The involvement of external providers in accordance with this additional agreement is permitted insofar as this is necessary for the provision of the additional functions. Use by external providers for their own purposes is excluded.
3.3 The special restrictions set out in sections 4.2 and 5.3 apply to private user data.
3.4 The user is obliged to correctly declare AI-generated or AI-modified media (2.21) within UGC (2.10) using the labeling function provided for this purpose. If this function is not available in the relevant area, the labeling must be replaced by a notice that is clearly recognizable to other users – either directly on the image/video (e.g., overlay/watermark) or in the associated post/comment. The labeling must be truthful, clear, and media-related.
4. Internal processing – purposes and limits
4.1 We may process user-generated content (UGC) within the scope of Section 3 on internal AI systems for the additional functions described in Section 2.12.
4.2 Private user data will only be processed if this is absolutely necessary for one of the following purposes:
- (a) for security purposes (e.g., spam and fraud prevention in private messages),
- (b) to provide functions you use (e.g., displaying products on watch lists), or
- (c) for additional functions you actively trigger (e.g., summaries or AI-supported help functions within private areas).
4.3 Before any processing of content within the scope of the additional functions described in Section 2.12, the PII gate defined in Section 2.15 is used; it is intended to prevent unauthorized external processing/transfer of personal data in accordance with Sections 5.1 and 5.3 and, in addition, to limit external transfer to the extent permitted under Section 5.1. Detection is performed automatically and is not error-free.
4.4 Regardless of the opt-out settings regulated in Section 6, content and security checks—including automated, AI-supported procedures where applicable—remain active at all times. They serve exclusively to maintain platform security, legal compliance, and community protection and cannot be deactivated.
4.5 We may analyze usage data (2.20) on internal AI systems, convert it into embeddings/vector representations, and—to the extent necessary for the purposes specified in Section 2.12 (in particular personalization, search/sorting, quality assurance, A/B testing, and abuse prevention) – train or fine-tune our own models. You may object to the use of your usage data for training/fine-tuning internal models in accordance with Section 6.5 with effect for the future; Section 6.2 (Effects on quality) and Section 6.3 (Future effect/model parameters) apply accordingly.
4.6 We may train and fine-tune internal AI systems with UGC to the extent necessary for the additional functions specified in Section 2.12 (in particular search/sorting, personalization, quality assurance, and abuse prevention). In doing so, we use the PII gate in accordance with Section 4.3. You can object to this use at any time via your account settings (Section 6.1); opt-outs only apply to the future; existing model parameters remain unaffected (Section 6.3). External training with UGC does not take place; Section 5.2 applies.
5. External providers – Permissible scope
5.1 External providers may perform AI evaluations for UGC as processors bound by instructions in order to provide or improve the additional functions described in Section 2.12, including, but not limited to, translations and content moderation. Use for the external providers' own purposes (in particular, their own training or profile creation) is excluded. PII may be processed in this context; processing shall be carried out exclusively for the specified purpose, in accordance with our instructions and within the framework of a data processing agreement.
5.2 External training or fine-tuning with UGC or private user data does not take place as a matter of principle. If the use of external providers becomes necessary for the development or operation of individual additional functions, processing for training or fine-tuning purposes may only take place:
- (a) to the extent that it is technically necessary;
- (b) exclusively on anonymized (Section 2.13) or pseudonymized (Section 2.14) data;
- (c) by processors bound by instructions;
- (d) after prior general information to users by means of an appropriate advance notice; individual consent is not required; and
- (e) for models that are operated and used exclusively on behalf of and for the purposes of Crazypatterns.
5.3 Private user data is generally not processed externally. Processing by external providers is only permitted if it is absolutely necessary or serves exclusively to provide a function that you use. Processing is always carried out for a specific purpose and under our instructions by processors bound by our instructions.
5.4 External providers may, as processors bound by instructions, process usage data (2.20) based on inference and embeddings (including vectorization/indexing) for the purpose of providing or improving the additional functions described in section 2.12. Model training/fine-tuning based on usage data by external providers is only permitted under the strict conditions of section 5.2 (a)–(e). Use for the providers' own purposes (in particular, their own training or profiling) is excluded; Section 5.1 applies accordingly. If usage data contains personal data, the PII gate (2.15) applies prior to external processing; Section 6.4 applies accordingly.
6. User control options
6.1 The user can object to the use of their UGC for training or fine-tuning on internal AI systems via the settings in their user account (opt-out).
6.2 The user acknowledges that opting out may impair the quality of personalized recommendations and other platform functions.
6.3 Opt-outs only apply to the future. Technical processing that has already taken place and existing model parameters remain unaffected.
6.4 The PII gate defined in Section 2.15 is taken into account when implementing the opt-out settings and applied accordingly.
6.5 Users can object to the use of their usage data (2.20) for training or fine-tuning internal AI systems via the settings in their user account. Opt-outs only apply to the future; existing model parameters remain unaffected (section 6.3). Note: An opt-out may reduce the quality of personalized content and recommendations (section 6.2).
7. Final provisions
7.1 In all other respects, the User Terms and Conditions apply.
7.2 Changes to this supplementary agreement shall be made in accordance with the mechanism provided for in the User Terms and Conditions.
8. Liability for AI functions
8.1 The liability provisions of the User Terms and Conditions (Section 13) apply accordingly to the use of AI functions.
8.2 AI-supported results (e.g., summaries, recommendations, translations, categorizations, scores) may be incomplete or incorrect in terms of content and do not constitute a representation or warranty; the respective original content, if available, remains decisive.
8.3 Sorting, visibility, and recommendations may change due to model updates, tests, or measures to prevent abuse and do not justify any claim to a specific display or reach.
8.4 The availability of AI functions may vary for technical reasons (e.g., maintenance, peak loads, third-party disruptions). There is no entitlement to uninterrupted provision.
8.5 No exclusively automated decisions with legal effect or similar significant impairment are made about users.
8.6 In the event of failures or errors on the part of processors used, we are only liable within the scope of the liability provisions of the User Terms and Conditions (Section 13).
Module B – Additional provisions for authors
9. Validity, inclusion, priority
9.1 This module applies in addition to Module A for users who upgrade their user account to an author account.
9.2 In the event of discrepancies, the provisions of this module shall take precedence for the matters regulated herein.
9.3 Authors can control certain AI processing (including PII) via their author account in accordance with Section 2.8 using opt-out settings.
9.4 The author warrants that they hold all rights to all product presentation data (2.5) and digital products (2.6) provided by them that are necessary for the processing provided for in this supplementary agreement. The author further warrants that all necessary third-party rights have been lawfully obtained and effectively granted for these processing operations, and undertakes to provide us with appropriate evidence upon request.
10. Grant of rights for product presentation data
10.1 The general rights of use are set out in Section 5 of the Author's Terms and Conditions.
10.2 In addition, you grant us permission to use product presentation data within the scope of the additional functions, in particular for automated analyses (AI evaluations), including the creation and storage of feature or vector representations and their use for indexing and findability within the platform.
10.3 The involvement of external providers in accordance with Section 5.1 is permitted insofar as this is necessary for the provision of the additional functions.
10.4 For digital products (2.6), the special restrictions under Section 11 apply.
11. Digital products – Special restrictions
11.1 Digital products shall not be processed externally or used for training or fine-tuning. Notwithstanding this, product-related external processing is only permitted in accordance with Section 11.4.
11.2 Unless expressly deactivated by the author via the settings in the author account, only the creation of summaries (2.7) and the feature or vector representations required for this on internal AI systems is permitted. These summaries may be processed externally, provided that they do not allow any conclusions to be drawn about the complete content and are used exclusively for functional processing within the platform and its official communication and distribution channels (e.g., website, newsletter, social media).
11.3 The author may deactivate the creation of summaries for their digital products account-wide; Section 12.6 remains unaffected.
11.4 The author may activate/trigger external processing for individual digital products (Section 2.6) for the use of additional extended functions – e.g., translation, spelling/grammar check, format/quality check – on a product-specific and explicit basis via the author account ("at the touch of a button"). Processing is carried out on a one-time or ad hoc basis and for a specific purpose as processing by a Processor under our instructions, without own use/training, with deletion of the transmitted files after service provision or in accordance with the contractual deletion periods of the respective service.
12. Control options and obligations of the author
12.1 Settings can be made account-wide.
12.2 The following processing operations can be optionally activated/deactivated:
- (a) internal inference and internal creation of vector/feature representations;
- (b) external inference and external creation of vector/feature representations.
12.3 The author may opt out of training/fine-tuning on internal AI systems with regard to their product presentation data at any time.
12.4 If the author has content (see 2.5 and 2.6) for which there is a license prohibition or insufficient rights for processing by AI systems, they are obliged to opt out of this content in accordance with Section 2.8.
12.5 Deactivating AI functions may affect findability, visibility, sorting, and recommendations, as well as automatically provided help functions, thereby reducing the reach and sales success of products.
12.6 Opt-outs in accordance with the above provisions shall only apply to the future. Technical processing that has already taken place and existing model parameters shall remain unaffected; affected content shall be excluded from future training runs.
13. Labeling of presentation media
13.1 The author is obligated to label AI-generated or AI-modified media (2.21) within their product presentation data (2.5) posted by the author on the platform itself and the media components of digital products (2.6) using the labeling function provided for this purpose. If this function is not available in the relevant platform area, the author must label the media by means of a notice that is clearly recognizable to users – either
- (i) in text form in the respective file of the digital product (2.6), or
- (ii) directly on the image/video in question, or
- (iii) in the text or comment field associated with the image/video. The labeling must be truthful, clear, and media-specific.
13.2 We are entitled to check compliance on a random basis and, if there are specific indications, to request suitable evidence (e.g., source files, editing logs) to the extent necessary and reasonable; further contractual and legal claims remain unaffected.
13.3 In the event of violations, we are entitled to label, downgrade, temporarily remove offers, or restrict the author account.
14. Compensation
14.1 The rights granted under this module are fully compensated by the remuneration in accordance with the Author Terms and Conditions.
15. Final provision for Module B
15.1 In all other respects, Module A and the Author Terms and Conditions apply.