OmevAI AI and Customer Content Terms

These AI and Customer Content Terms form part of the OmevAI Terms of Service. Capitalised terms not defined here have the meanings given in the Terms.

Mandatory training notice: OmevAI always uses Customer Content, including inputs and outputs, to train, evaluate, test and improve models and related service systems. There is no customer opt-out. Do not submit Customer Content if you do not authorise this use.

1. What is Customer Content?

Customer Content includes:

prompts, messages, instructions, keywords and parameters;
text, images, documents, URLs, snippets and other uploaded or transmitted material;
data imported from a connector or integration at your direction;
saved project inputs, configurations and results; and
responses and other outputs generated for you through the Service.

Account registration details, billing information held by Creem, and technical security/usage logs are governed by the Privacy Policy and are not Customer Content merely because they relate to you.

2. Your inputs and outputs

You retain the rights you already hold in your inputs. As between you and OmevAI, you own an output to the extent applicable law recognises copyright or another protectable right in that output.

AI output may not qualify for intellectual-property protection and is not guaranteed to be original or exclusive. Other customers may receive similar or identical output. OmevAI does not transfer rights that belong to another person and does not warrant that an output is non-infringing.

You remain responsible for reviewing, editing, fact-checking and clearing an output before using or publishing it.

3. Licence to OmevAI

You grant OmevAI a worldwide, non-exclusive, royalty-free, transferable and sublicensable licence to host, reproduce, transmit, analyse, annotate, adapt, modify and otherwise use Customer Content:

to provide, secure, support and troubleshoot the Service;
to classify requests and select, operate and evaluate service providers;
to detect abuse, quality defects and security incidents;
to train, fine-tune, evaluate, test, benchmark and improve models and related systems; and
to create statistics, features, datasets and other derived materials consistent with the agreement.

The licence permits disclosure to personnel, contractors and service providers that support those purposes. It also permits use after account closure within the retention and survival rules below.

You receive no royalty or other compensation for model or service improvements created using Customer Content.

4. Mandatory nature of model improvement

Model training and improvement are a material part of the OmevAI service model, not an optional marketing permission. You cannot disable them for an account, project, key, request or integration.

If you object to this use as a Customer, your remedy is not to purchase or use the Service and not to submit Customer Content. This contractual rule does not waive rights that an individual data subject has under a mandatory data-protection law; OmevAI will assess a valid statutory request under the Privacy Policy.

5. Customer representations

For every item of Customer Content, you represent and warrant that:

you own it or have all permissions necessary for every use described in the agreement;
the content and your instructions comply with law and the Acceptable Use Policy;
you have provided legally required notices and obtained any legally required permissions;
it does not breach a duty of confidence or restriction owed to a third party; and
if Enterprise personal-data processing has been approved, you have a lawful basis to disclose the agreed data for service delivery and OmevAI's separately disclosed model-improvement purpose.

Do not submit material under a confidentiality, data-protection or licensing obligation that is incompatible with mandatory training and the licence above.

6. Self-service personal-data prohibition

Self-service Customer Content must not contain information relating to an identified or identifiable natural person. This applies whether the individual is you or someone else and whether the information appears sensitive or ordinary.

Removing a name is not enough if a person can still reasonably be identified from identifiers, context, combinations of attributes or other available information. Pseudonymised or indirectly re-identifiable information remains personal data.

Genuinely anonymous statistics, aggregate information and synthetic examples are permitted if no person is reasonably identifiable and the information cannot reasonably be linked back to an individual.

7. Enterprise personal-data processing

An Enterprise customer may submit agreed categories of ordinary personal data only after:

direct manual review of the use case by OmevAI;
execution of a confidentiality agreement;
execution of a Data Processing Addendum containing the named subprocessor annex; and
written activation of that specific personal-data-enabled use case.

The DPA applies when OmevAI processes personal data solely on the Enterprise customer's documented instructions to deliver the Service. When OmevAI determines the separate purpose and essential means of general model training, evaluation and improvement, OmevAI acts as an independent controller for that processing. That controller activity is mandatory and is not converted into instruction-only processing by the DPA.

Enterprise approval is specific and does not authorise the prohibited categories below.

8. Prohibited identifiable sensitive and high-risk data

You must not submit Customer Content that relates directly or indirectly to an identified or identifiable natural person and includes:

racial or ethnic origin, political opinions, religious or philosophical beliefs, or trade-union membership;
genetic data;
biometric data used or capable of being used to uniquely identify a person;
physical or mental health, diagnosis, treatment or disability information;
sex life or sexual orientation;
criminal allegations, convictions or offences;
data about a person under 18;
government-issued identifiers or copies of identity documents;
precise location, private communications or comparably sensitive personal information; or
any personal data that you lack authority or a lawful basis to process and disclose for all stated OmevAI purposes.

This restriction is about identifiable people, not general subject matter. Anonymous medical statistics, synthetic patient scenarios, non-identifying genetic research, and other genuinely anonymous or aggregate material may be submitted if no person is reasonably identifiable.

9. Absolute prohibition on credentials and payment instruments

Do not submit any real:

payment-card number, bank-account credential, payment token or other usable payment instrument;
password, one-time code, recovery code or authentication secret;
API key, private key, access token, session token or signing secret; or
comparable credential capable of authorising access, payment or a transaction.

This prohibition applies even to corporate accounts and even if the information is not personal data. Non-functional synthetic, masked or irreversibly tokenised examples are permitted only when they cannot be used or restored as real credentials or payment instruments.

10. Provider processing

OmevAI may send Customer Content to one or more AI inference and infrastructure providers to generate, validate, route or improve responses. A request may be processed in the EEA or in other regions, including the United States and Asia-Pacific, as described in Subprocessors and Data Transfers.

Self-service customers do not receive confidential details of models, routing sequence, vendor economics or internal quality controls. Enterprise customers approved for personal-data processing receive the legal identities and processing locations of potential subprocessors in the confidential DPA annex before signing it.

11. Retention and deletion

OmevAI may retain identifiable raw Customer Content, including prompts, responses and copies selected into training or evaluation datasets, for no longer than 24 months from collection.

At the end of that period, OmevAI will delete the source content or irreversibly anonymise it so that no individual is reasonably identifiable and the source content cannot reasonably be reconstructed.

OmevAI may retain irreversibly anonymised data, aggregate statistics, evaluation metrics and trained model parameters until the relevant model or system is retired, provided the original personal data or source content cannot reasonably be extracted or reconstructed from them.

Account closure or contract termination does not require OmevAI to unwind completed training, delete compliant aggregate information or destroy non-extractable trained parameters. Legal holds, fraud investigations and mandatory recordkeeping may temporarily delay deletion of a limited copy, which will be isolated and used only for that purpose.