Claude Code Privacy Flaw Exposes User Emails in Web Requests

Privacy Breach: Claude Code Leaks User Emails in HTTP Headers

In a concerning development for developers using Anthropic's Claude Code, a privacy vulnerability has been exposed that sends users' email addresses to arbitrary websites during web searches. The issue, reported on GitHub and discussed on Hacker News, reveals that the AI coding assistant embeds the user's email in the User-Agent request header, potentially exposing sensitive information to third parties without consent.

The report, filed by user taylor-edwards on November 5, 2026, details how this behavior led to a flood of scam emails. "I receive dozens of scam emails per day because claude code divulged my email to arbitrary 3rd parties in the User-Agent request header while conducting web searches without my permission," the report states. This is not an isolated incident; other users on Reddit have reported similar issues, including one where Claude Code sent a personal email to SEC.gov.

The Technical Flaw: How Email Addresses End Up in Headers

The root cause appears to be that Claude Code includes the user's account email in its system context, which is then used to construct the User-Agent header. According to the Reddit post, "Claude Code includes your account email in its system context," and this information is inadvertently transmitted when the tool makes HTTP requests to fetch web content or interact with APIs.

The User-Agent header is a standard part of HTTP requests, typically identifying the client software. However, embedding a personal email address in it is a serious privacy violation, as it allows any website visited to harvest the email without the user's explicit knowledge or consent. This is particularly dangerous for developers who use Claude Code for web scraping or automated browsing, as they may unknowingly expose their email to countless domains.

Anthropic's Response: The 'cc-plugin-privacy' Mod

Anthropic has responded to the issue by introducing a built-in mod called cc-plugin-privacy in version 2.1.317 and later. According to daily.dev, this mod "deterministically blocks this behavior" and is enabled by default. Users can disable it via the /plugins command, but the default is to have it active.

However, this fix has been met with skepticism. The daily.dev community analysis shows that 75% of commenters are skeptical of the solution. Critics argue that making it an opt-out plugin rather than removing the data entirely is inadequate. The original GitHub issue demands that the behavior be "removed from the harness entirely," stating that it should not be a setting with opt-in or opt-out behavior.

Community Reaction: Trust Eroded

The developer community's reaction has been largely negative. On Hacker News, user matheusmoreira commented, "That is just disgusting. So happy I switched away from Anthropic." Another user, EGreg, expressed a preference for using chatbots on websites instead of Claude Code, saying, "I've never used Claude Code or Cursor. All the code I got written this year was without them."

The frustration is palpable, as users feel their privacy has been compromised by a tool they trusted. The fact that the email is sent without explicit permission, and that the fix is not a complete removal, has led to a loss of faith. The original reporter stated they are "migrating my company to Pi with OpenRouter since I have no faith in anthropic preserving my privacy."

Why This Matters: Privacy in AI Tools

This incident highlights a growing concern in the AI coding tool space: the handling of user data. As AI assistants become more integrated into development workflows, they have access to sensitive information, including emails, API keys, and code. Ensuring that this data is not inadvertently exposed is critical for maintaining user trust and security.

The issue also raises questions about the design of AI tools. Why would an AI coding assistant include an email in a User-Agent header? It's likely intended for identification or debugging, but it demonstrates a lack of consideration for privacy. The fact that it was not caught during development suggests a need for better privacy audits in AI tool development.

Moreover, the fix itself is a compromise. While the cc-plugin-privacy mod blocks the behavior by default, it is still a plugin that can be disabled. This leaves the door open for users who unknowingly disable it, or for future versions to regress. The community's demand for a complete removal of the email from headers is a reasonable request that Anthropic should consider.

Conclusion: A Call for Better Privacy Practices

Claude Code's email leakage is a stark reminder that AI tools must prioritize user privacy by design. While Anthropic has taken a step to mitigate the issue, the response has not fully addressed the community's concerns. As AI coding assistants become more prevalent, developers must be vigilant about the data these tools collect and transmit.

For now, users of Claude Code should ensure they are on version 2.1.317 or later to benefit from the cc-plugin-privacy mod, and they should be cautious about the information they share with the tool. Anthropic, on its part, should consider making the removal of email from headers a permanent, non-negotiable feature, not just a default-on plugin. The trust of its user base depends on it.