According to GitHub issue #107757 filed November 5, Claude Code version 2.1.318 on macOS was writing signed-in users' email addresses into HTTP User-Agent headers during web requests without permission, leading one user to report receiving scam emails daily. Anthropic contributor Lydia Hallie announced a fix planned for early the following week that removes account email from the context provided to the model, and noted that a built-in privacy plugin blocks the leak in version 2.1.317 and later.
Topics #
Sources #
- Official[Read article](https://github.com/anthropics/claude-code/issues/107757)
- Press[Read article](https://theainewsreport.com/2026-11-06-claude-code-email-user-agent-leak-privacy-mod-explained.html)
Go deeper #
This intelligence is sourced automatically from public sources across the web and synthesised by the Prefactor AI pipeline. Stories are reviewed before publication.
source & further reading
getreadyforagents.com — original article
AJX launches open-source CLI and skill bundle for testing product usability from AI agent perspective
Apiiro reports Claude Code leaking secrets via 'Don't Ask Again' feature
Plannotator releases inbox interface for human review and annotation of AI agent work outputs