Sanitize CSV for ChatGPT & Claude
Redact sensitive spreadsheet columns or replace identifiers with consistent keyed pseudonyms locally. Preserve useful relationships while reviewing indirect identifiers, headers, and retained columns.
Redact PDF, Table, or Image
PDF · CSV · TSV · XLSX · PNG · JPEG · WebP · BMP
Pseudonymize Customer CSV Data Before AI Analysis
Customer names and emails are not the only sensitive fields in a table. Order notes, exact locations, employee roles, and rare purchase combinations can identify a person after direct identifiers are removed. Choose column actions deliberately and review the sanitized preview before deciding what to share.
Consistent Keyed Pseudonyms
Keyed Hash uses HMAC-SHA-256 with a locally held key and a 128-bit exported identifier. Repeated values with the same key receive the same pseudonym. This preserves record relationships, not anonymity; protect the key and assess remaining identifying information.
Synthetic Identifier Replacement
Replace repeated values within a column with labels such as Record_0001 and Record_0002. These labels are consistent within that column and processed file, not across separate files. Column headers remain unchanged and must also be reviewed.
Sanitize TSV and Headerless CSV Files
Delimiter detection supports CSV and TSV input, including quoted fields and multiline cells. For headerless data, disable the header setting so the first row is sanitized too. XLSX workbooks up to 10 MB open locally with a worksheet selector. Export the selected sheet as sanitized CSV values. Formulas are not recalculated; review dates, cached formula values and number formatting.
Questions about this workflow
How can I sanitize a CSV before sending it to ChatGPT or Claude?
Choose a local CSV or TSV, review every column action, and preview the result. Replace sensitive columns with redacted values, keyed hashes, or consistent synthetic IDs. Download the sanitized table and review it before independently sharing it with an AI service. This tool does not send data to ChatGPT or Claude.
How do consistent CSV pseudonyms preserve repeated customer records?
Keyed Hash uses HMAC-SHA-256 and exports a 128-bit identifier. Equal non-empty values with the same key receive the same pseudonym. Synthetic IDs such as Record_0001 preserve repeated values within each column of one processed file; they are not stable identifiers across separate files.
Can I sanitize TSV files or CSV data without a header row?
Yes. The parser detects supported delimiters and preserves the delimiter on export. Turn off First row contains headers for a headerless table so its first row is processed as data. Review all column actions after changing this setting.
Does masking or hashing a CSV make it anonymous?
No. Partial masking retains fragments, keyed pseudonyms preserve relationships, and kept columns may allow re-identification. Headers can also contain sensitive text and remain visible. Review combinations of values, not just obvious names or emails.
Can I sanitize an XLSX worksheet before sharing it with an AI?
Yes. Open an XLSX workbook up to 10 MB, select the worksheet, apply column rules and export sanitized CSV values. Dates and numbers use their stored formatting; formulas are not recalculated. Other sheets, formatting and formulas are not included in the CSV. No AI upload integration is included.