This is an English translation provided for convenience. The legally binding text is the Russian version at flyai.by/privacy. In case of any discrepancy, the Russian version prevails.
This Personal Data Processing Policy (the “Policy”) has been prepared in accordance with the Law of the Republic of Belarus No. 99-Z of May 7, 2021 “On Personal Data Protection” (the “Law”) and the Google API Services User Data Policy.
Limited Liability Company «AI Solutions» (in Russian: ООО «ИИ Решения»; registration number (UNP) 194000696, Republic of Belarus), the owner and operator of the flyai.by corporate AI infrastructure platform (the “Platform”, the “Operator”), applies privacy by design: the data we collect is limited to what the Platform cannot work without, and protection is ensured by how the system is built rather than by a promise.
1. Operator and responsible person
- Operator: Limited Liability Company «AI Solutions», UNP 194000696.
- Address: Lyubimova Ave. 39, premises 1N, Minsk 220117, Republic of Belarus.
- Person responsible for internal control over personal data processing: Director Mikalai Belchyn.
- Contact for personal data matters: info@flyai.by, phone +375 29 503 43 91.
This Policy describes the processing of data for which FlyAI is the controller. Where personal data controlled by the customer itself passes through the Platform — data of its employees, counterparties or clients — the roles and obligations are set out in the Data Processing Agreement (in Russian): in that case the customer is the controller and we are the processor acting on its instructions.
2. Categories of data, purposes and legal bases
The Platform is intended for business customers. As a matter of principle, we do not store or process the personal data of payers who pay through the ERIP payment system: no surname or name, no address, no phone number. Only the fact of payment is recorded — invoice number, amount, date and status. The bank already holds this data, and a second copy with us would add risk without benefit.
| Purpose | Categories of data | Legal basis | Retention period |
|---|---|---|---|
| Operating the customer account | Email address, password (as an irreversible hash), customer company name, role in the workspace | Contract with the data subject or their employer (Art. 6(15) of the Law) | Term of the contract and 3 years after its termination |
| Access security (at the user’s choice) | Two-factor authentication data, Telegram chat ID | Consent of the data subject (Art. 5 of the Law) | Until the feature is disabled or consent is withdrawn |
| Processing requests by the AI partner | Text of requests and the context sent with them | Contract with the data subject or their employer | Stored encrypted until the contract is terminated |
| Payment confirmation and usage limits | Invoice number, amount, date, payment status | Contract with the data subject or their employer | Periods set by accounting legislation |
| Connecting external services (at the customer’s choice) | Address of the connected account, encrypted refresh token | Consent of the data subject expressed in the external service’s interface | Until the customer revokes access |
3. Protecting conversations: what we technically cannot do
Records of interactions with the AI partner are encrypted on the server with the public half of a cryptographic key. The private half, which is required for decryption, is not present in the Operator’s infrastructure — it is kept on an external device by the key owner. The Operator has no technical ability to read the contents of conversations: this is a property of the system design, not a commitment to refrain.
Before a request is sent outside the protected perimeter, personal data is automatically removed from it according to rules configured in the customer’s workspace. The exact composition of data leaving the perimeter is recorded separately, to answer lawful requests from competent authorities.
A caveat we state openly: while generating a response, the AI partner processes the request text in plain form in operating memory — otherwise processing is impossible. The plain text is not written to disk and does not leave the perimeter.
Where the data physically resides. The Platform’s information resources are located in the Republic of Belarus. The infrastructure provider is engaged by us as a processor on terms no weaker than those we accept towards the customer and has no right to use the data for its own purposes. Only requests to artificial intelligence models leave the country — as described in the next section.
A consequence the customer should know in advance: if all keys of a workspace are lost, the conversation history cannot be restored by the Operator or anyone else. There is no “recovery through support” procedure — if there were, the protection itself would not exist.
4. Cross-border data transfer
The AI partner requires the computing capacity of foreign providers of artificial intelligence models. This means that masked requests are transmitted over secure channels to servers outside the Republic of Belarus.
This computing capacity may be located in countries that are not on the list of countries ensuring an adequate level of protection of data subjects’ rights. Cross-border transfer is therefore carried out on the basis of the data subject’s informed consent (Art. 9(1), paragraph 3 of the Law).
Risks we are obliged to warn about. Countries that do not ensure an adequate level of protection may lack independent data protection authorities, the scope of data subject rights may be narrower than under the legislation of the Republic of Belarus, and access by public authorities to data may be broader. By starting to use the Platform, you confirm that you have been informed of these risks and consent to the cross-border transfer of your requests.
The names of specific computing capacity providers are the Operator’s trade secret. The Operator’s contracts with these providers prohibit using the transferred data for training or fine-tuning models. The Operator cannot be responsible for the actions of third parties beyond the terms of these contracts and makes no representations beyond them.
5. Connecting Google Workspace
Connecting Google is an optional feature enabled by the customer. Access is granted by the user themselves in Google’s interface; the Operator receives a refresh token, which is stored sealed in a closed container, is never transmitted over the network to the outside and is decrypted only in operating memory at the moment an operation is performed. The address of the connected account is also stored, so that the customer can see in their account what exactly is connected.
5.1. Platform application mode
By default, the Platform’s application is used, which requests a single access scope: https://www.googleapis.com/auth/drive.file. It grants access only to files created by the application itself. The user’s existing files, folders and documents are not accessible to the Platform. The access scope cannot be extended in this mode.
5.2. Customer application mode
The customer may connect its own Google application registered in its organization by entering its client ID and secret in the account. In this case, access to the customer’s data is granted to the customer’s own application, and the Operator acts on the customer’s instructions.
In this mode the customer may choose an extended access scope — read-only access to Google Drive files (https://www.googleapis.com/auth/drive.readonly). When the extended scope is chosen, the following rules apply, and the Operator follows them regardless of the scope of access granted:
- the customer’s files are not copied into the Platform’s storage and are not indexed on its side;
- files are read only upon a specific user action during a conversation and only for the time needed to generate a response;
- every access to files is recorded in a log available to the customer for review;
- no automatic crawling, inventory or background scanning of Drive contents is performed.
5.3. Limited Use
The flyai.by Platform’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.
- No model training. Data obtained from Google Workspace is never used to train, fine-tune or develop artificial intelligence or machine learning models — neither generalized nor personalized ones.
- No human access. The Operator’s staff do not view the user’s Google files, except with the user’s explicit consent for technical support, to investigate security incidents, or as required by law.
- No transfer to third parties. Google data is not sold to third parties and is not used for advertising or targeting.
- Disconnection and deletion. The user may revoke the Platform’s access with one action in the account and in their Google account settings. The stored address and encrypted token are then deleted.
6. Rights of data subjects
Under the Law, you have the right to:
- withdraw your consent at any time without giving reasons;
- obtain information about which of your data is processed and to whom it is transferred;
- demand that data be changed or deleted where there are grounds.
To exercise these rights, use the account at cabinet.flyai.by or send an application to the Operator’s address, or as an electronic document signed with an electronic digital signature to info@flyai.by. The application must contain your surname, first name and patronymic, address of residence (stay), date of birth, identification number (or, if none, the number of your identity document), the substance of your requests, and your handwritten or electronic digital signature. The review period is 15 days.
If you disagree with the Operator’s actions, you may file a complaint with the National Personal Data Protection Center of the Republic of Belarus.
7. Changes to the Policy
The Operator may change this Policy. A new edition takes effect when published at flyai.by/privacy. Previous editions are kept and provided on request at info@flyai.by; the edition in force when a particular customer accepted the terms is recorded by the Platform.