How Clare protects your identity
Your messages are filtered before they reach the AI. Names, places, and contact details never reach Clare as written.
What gets replaced
Clare uses automatic detection to find two categories of personal information in your messages.
Names, locations, and group affiliations
These are replaced with consistent anonymous placeholders. The same name always maps to the same placeholder throughout a session, so Clare can follow the story without ever knowing who anyone really is.
- Personal names
- Cities, countries, and other geographical locations
- Nationalities, religious groups, and political affiliations
Contact and financial identifiers
These are removed entirely, because they carry no meaning relevant to the conversation.
- Email addresses
- Phone numbers
- IP addresses
- Bank account numbers (IBAN)
- Credit card numbers
Example
Here is what a message looks like before and after processing.
My manager Lena moved our whole team to Berlin last year. Since then our colleague Marco has been harder to reach. You can try him at marco.b@company.com or +49 30 1234 5678.
My manager [Person A] moved our whole team to [City A] last year. Since then our colleague [Person B] has been harder to reach. You can try him at ***EMAIL_ADDRESS*** or ***PHONE_NUMBER***.
Placeholders are consistent
If you mention the same person twice, both mentions become the same placeholder. Clare can follow the conversation coherently without ever learning the real name.
See it for yourself
Inside the chat interface, you can open the What Clare sees panel at any time. It shows you exactly which version of your messages was sent to the AI, after all replacements have been applied.
Limits of automatic detection
Automatic detection works well in most cases but is not perfect. Unusual name spellings, abbreviations, or indirect references may occasionally be missed. For full control over what you share, avoid mentioning identifying details that are not relevant to the situation.