[BUG] Claude Code claiming false Usage Policy violation when scanning emails

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

This is the same issue as was already reported months ago: #18053, that issue was closed without resolution. Like the other poster, I too ran into the same error while scanning emails. Coincidentally this was also scanning HARO feeds, just like the other user. I'm wondering if the issue is due to Claude misinterpreting content within those emails as instructions, or some poison pill within the content:

<img width="942" height="204" alt="Image" src="https://github.com/user-attachments/assets/292a9329-502c-4ce9-8989-e312295f8c0a" />

What Should Happen?

Claude should recognize that the violation is not caused by the user but by specific content in the email, it should explain what caused the violation and present user with sane options:

1. process the content differently
2. skip malformed content

Error Messages/Logs

API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy (https://www.anthropic.com/legal/aup). Please double press esc to edit your last message or start a new session for Claude Code to assist with a different task. If you are seeing this refusal repeatedly, try running /model claude-sonnet-4-20250514 to switch models.

Steps to Reproduce

1. sign up for HARO digest in your mailbox: https://www.helpareporter.com
2. setup gmail MCP to read emails
3. ask claude to read/profile/summarize emails, eventually (after a few days/weeks of digests), you will encounter this issue

Claude Model

Opus

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

2.1.123

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Other

Additional Information

_No response_