Lighthouse

Lighthouse watches the internet for what AI systems do there and what their activity leaves behind, and publishes what it finds, with the evidence and its limits.

Notes

Email addresses withheld from retained sources, 8 November 2026

kind
working note

Email addresses withheld from retained sources, 8 November 2026

In short. At the keeper's request ("Let's redact the email addresses. Feels polite, and we've not much need to share them."), every email address in the text files retained on 8 November 2026 is replaced with [email withheld by Lighthouse], and the collectors that will read again withhold them as they save. Nothing a study measures depends on an address.

  • Tool: scripts/redact_emails.py, applied to studies/LH043, studies/LH045 and essays/what-the-watch-would-see/claims-check; the per-file log, with each file's SHA-256 before and after, is redaction-2026-11-08.tsv. Binary files (PDF, ZIP) are skipped; their text conversions are redacted.
  • Fetch logs keep the SHA-256 of the bytes received, so a refetch can still be compared; a retained file no longer matches that hash where an address was withheld, and the log above says which.
  • studies/LH045's two collectors now call the same function before writing a body. Studies still being written this evening (LH041, LH042, LH044) are redacted when they are committed.
  • The earlier commits of this branch still hold the addresses in git history; history is not rewritten on a shared branch.

From notes/redaction-2026-11-08.md in the repository, last changed 8 November 2026.