MIRA Integrity
Manuscript Integrity & Reference AnalyzerHow we handle your manuscript
Data retention policy — applies to the closed beta of MIRA (web version).
In short: Your manuscript is deleted immediately after processing. Reports are automatically deleted after 30 days. No manuscript content is retained for research, validation, or system improvement without your explicit consent. (Text processed by Anthropic is deleted by Anthropic within 30 days; see section 6.)
1. Your manuscript
The manuscript file you upload (.docx) is deleted immediately after MIRA has processed it. While it waits for processing, it is kept only in temporary storage that is not backed up, under a random name; its original file name is not stored.
2. Working files
While checking your manuscript, MIRA creates working files: the extracted reference list, the claims taken from your text, and the reference-check results. These contain text from your manuscript. They are deleted immediately after your report has been produced.
3. Your report
Your report contains the text of the claims that were checked, so we protect it in the same way as your manuscript. It is kept for 30 days so that you can come back to it or tell us about a problem. After 30 days it is deleted automatically. You can delete it earlier yourself, at any time, with Delete this report now on the report's status page. You can also write to beta@mira-integrity.com with your access code, and we will confirm when it has been deleted.
4. What we keep
We keep statistics and reliability measures (counts and statuses, such as how many references were found or not found, how often a check failed for technical reasons, and how long processing took) and technical logs (dates, times, identifiers and processing status). These contain no text from your manuscript. We use them to monitor and improve how reliably MIRA works, and we may keep them indefinitely.
5. Your consent comes first
- No manuscript content is used for research, validation studies, training, test sets or publications unless you explicitly agree.
- If you report a problem, you may choose to let us keep that one claim and its cited reference beyond 30 days, so that we can investigate and fix the problem. Only that item is kept, not your manuscript or the rest of your report. You can withdraw this consent at any time by email.
- If you report a problem without this consent, we keep only the claim number, the type of problem and the date. Any comment you write is deleted together with your report.
- Any wider use, for example including a whole document in a validation study, would need your separate, specific agreement.
6. Who else processes your text
- The text of your manuscript is processed by Anthropic's Claude API under Anthropic's commercial terms. Anthropic does not use API inputs or outputs to train its models, and deletes them within 30 days, unless content is flagged for a usage-policy violation (then kept for up to 2 years) or longer retention is required by law. Processing may take place outside the EU, including in the United States, under Anthropic's Data Processing Addendum with Standard Contractual Clauses.
- Bibliographic details from your reference list (titles, DOIs, author names) are sent to CrossRef and PubMed to verify the references.
- Your files are stored on servers in the European Union (Netherlands).
7. Who can see your manuscript and report
Only you, with your access code, and the MIRA beta operator. Your files are not shared with anyone else. The MIRA beta is operated by the developer of MIRA, contact beta@mira-integrity.com.
8. Questions and requests
Email beta@mira-integrity.com. Please include your access code.