Privacy policy.

1. Who and what this policy covers

Unassuming AI is the independently operated inference service at unassuming.ai. In this policy, “we” and “our” refer to the operator of that service.

This policy covers our public website and policy pages, the /chat interface, and the text inference API under /v1. Experimental tools under /rsi are outside this inference-service policy and may use additional model providers. A third-party application or marketplace that connects to our API has its own responsibilities and privacy policy.

2. Request retention, debugging, and training

We process the prompts, conversation messages, and generation settings you send in order to produce a response. We may also retain submitted requests and associated diagnostic information so that we can promptly reproduce failures, identify their causes, and verify fixes when bugs occur. Saved requests may include prompts, conversation messages, and generation settings. Conversation history may include earlier model outputs if your client sends them back in a later request.

• Debugging purpose: saved request content is used to investigate and fix bugs in the service, not to create training datasets.
• No model training: we do not use customer requests, prompts, or completions to train, fine-tune, or update model weights.
• Limited access: access to saved request content is limited to the operator and people authorized to investigate and resolve service issues.
• No content sale or advertising use: we do not sell API request or completion content or use it to target advertisements.

We keep request records only for as long as needed to investigate the relevant issue and verify its resolution, and delete them when they are no longer needed for that purpose. We do not promise zero-day retention for request content. You may contact us to request deletion of retained information, as described in section 7.

3. Operational metadata and retention

Our web server records operational information such as connecting IP addresses, timestamps, requested URLs, HTTP status codes, response sizes, browser user-agent strings, and referrers when supplied. Error logs may contain diagnostic messages and request metadata. We use these records to operate, troubleshoot, and protect the service.

Web access and error logs are configured for daily rotation, with up to 14 rotated log files retained. This is a rotation-based retention rule: empty logs do not rotate, so archived files can remain longer during periods without new entries. These web server logs are not configured to record API request or response bodies; request content retained for debugging is governed separately by section 2.

System and relay diagnostics record operational events such as startup, errors, and tunnel connections; their retention is managed by system storage limits. They are not configured as prompt or completion logs. Inference counters and timing statistics are also used to understand service performance.

If you contact us, we process the contact details and information you choose to provide to handle your request. Correspondence is kept for as long as reasonably needed to resolve the issue and maintain necessary business or legal records. You may request deletion of correspondence that no longer needs to be retained.

Avoid putting secrets or prompt content into URL query strings, because requested URLs can appear in access logs. API access keys should be supplied using the supported authentication headers.

4. Chat history and browser storage

The /chat interface stores its API access key, conversation history, and settings in your browser’s local storage. This lets the interface remember your session and conversations on that browser. Messages needed for each inference request are sent to the API when you submit them and may be retained for debugging as described in section 2. Browser-local history and server-side diagnostic request records are separate.

Delete individual conversations using the chat interface, or clear site data for unassuming.ai in your browser to remove all locally saved conversations, preferences, and access keys. Clearing browser history does not delete request records already retained on the server; contact us to request deletion of those records. Other people with access to the same browser profile may be able to access locally stored information.

The public landing and policy pages do not load third-party analytics, advertising scripts, or tracking cookies. The chat interface uses browser storage for the functionality described above.

5. Infrastructure, transfers, and disclosure

Our public HTTPS gateway is hosted on OVHcloud in Canada. Inference requests pass through that gateway and an encrypted tunnel to an operator-managed GPU backend on a separate host. Public API traffic uses HTTPS. Hosting and network providers process data as needed to deliver this infrastructure.

Messages sent to our contact email address are forwarded by Porkbun to an operator-managed Gmail inbox. Porkbun and Google process that correspondence to provide email delivery and storage. This email arrangement is separate from the inference API; API prompts and completions are not sent to these email providers as part of inference.

We do not offer a contractual country-specific data-residency guarantee through this public endpoint. Contact us for the current processing arrangement before using the service with a specific residency requirement.

If you access the service through a marketplace or another application, that service receives and processes your information under its own policies. We may also disclose information when reasonably necessary to comply with applicable law, respond to valid legal process, address abuse, or protect the service and the rights of others. We limit such disclosures to information relevant to the purpose.

6. Why we process information

We process inference content to deliver the response you request, retain request records when needed to reproduce and fix bugs, use operational metadata to maintain reliability and security, and handle correspondence to provide support and respond to privacy requests. Debugging does not include using your requests or completions as model training data. Where applicable law requires a legal basis, we rely on performance of the requested service, our legitimate interests in operating and securing it, or compliance with legal obligations, as appropriate to the processing.

We do not use this information for advertising profiles or automated decisions about access to employment, credit, insurance, or other similarly significant opportunities. Model-generated answers are outputs requested by users; they are not decisions made by us about an individual.

7. Your choices and rights

Depending on the law that applies to you, you may have rights to access, correct, delete, or obtain a copy of personal information; to restrict or object to certain processing; and to complain to the relevant data-protection authority. Contact us to make a request. We may need enough information to verify the request and identify relevant records.

You may ask us about request records retained for debugging or request their deletion. An approximate request time or request identifier, if available, can help us locate records; do not send API keys. Diagnostic records are not a guaranteed backup of your conversations, and records may already have been deleted. Browser-local history must be managed in the browser where it is stored. We do not knowingly seek personal information from children under 18; our service is intended for adults.

Security measures reduce risk but cannot guarantee that every system or transmission will be free of unauthorized access or disruption. Share only information you are authorized to process through the service.

8. Changes to this policy

We may update this policy when our service or data practices change. The current version and effective date will be published on this page. Material changes to how inference content is used will be disclosed before those changes take effect, using this website and, where appropriate, an available account or platform communication channel.

9. Contact

For privacy questions, data-rights requests, or concerns about this service, contact the operator of Unassuming AI at contact@unassuming.ai.

Please include enough context to identify your request, but do not send API keys or unnecessary sensitive information.