適正利用

1. Purpose

This Policy protects research users, the people whose material may appear in uploads, data sources and the OncoRadar service. It applies to website, chat, uploads, workspace, API, MCP, automation and generated-output use.

2. Uploads, identifiable material and medical misuse

Upload documents, images and DICOM imaging only through the upload flow, which encrypts the file and de-identifies it before any model or reviewer sees content. Upload only material you have the legal authority to process, including any ownership, license, institutional, ethics-board or consent basis your jurisdiction requires. Do not attach or paste identifiable patient material through the API, MCP or chat text to bypass the upload flow. Do not attempt to re-identify any person from de-identified material, to defeat the de-identification pipeline, or to link derivatives back to a person.

Material intended for individual patient care is out of scope. Do not represent OncoRadar as a medical device or use it to diagnose, triage, prescribe, dose, determine eligibility, or make autonomous patient-care decisions. Findings are research patterns, not diagnoses.

3. Illegal, harmful or deceptive conduct

Do not use the service for unlawful surveillance, discrimination, harassment, exploitation, fraud, impersonation, misinformation presented as verified clinical fact, infringement, malware, credential theft, spam, phishing, or development of biological harm. Do not use generated outputs to conceal their automated origin where disclosure is required.

4. System and access abuse

Do not probe or exploit vulnerabilities without written authorization; access another tenant; scrape in a manner that violates source rights; bypass authentication, quotas, rate limits, upload limits or safety controls; upload executables, archives designed to exhaust processing, or malware; reverse engineer credentials; overload the service; or resell access unless an agreement expressly permits it.

5. Research and content responsibilities

Respect publication licenses, robots directives, contractual restrictions and citation requirements. Upload only content you may lawfully process. Independently validate AI outputs and do not fabricate citations, study findings or provenance. Contribute material to the research dataset only when you hold the authority the Dataset Contribution Terms require.

6. Enforcement and reporting

We may investigate, preserve relevant records, rate-limit, remove content, quarantine or purge uploads, revoke credentials or suspend access proportionately, including immediately for security risk. Report suspected abuse to abuse@oncoradar.org; security vulnerabilities belong at security@oncoradar.org. Good-faith reports do not authorize access beyond what is necessary to identify the issue.