[Bug] Anthropic API Error: False-positive cyber safeguard classification on harmless messages

Bug Description
Claude Code is incorrectly triggering the cyber safeguard on completely harmless messages. The issue reproduces in a fresh session and a completely separate empty directory. The message: "hii" is immediately classified as: [cyber] Opus 5 is blocked and automatically switches to Opus 4.8, which is then also blocked by the cyber safeguard. This is not related to the project, CLAUDE.md, project memory, or an existing conversation. I also updated Claude Code to version 2.1.280 and reproduced the issue afterward. Request ID from the blocked request: req_011CfJwmg5C7q83vt1AW4F5T Please investigate this as a false-positive safeguard classification on my account/session. I want to continue using Opus and do not want to disable or delete my Claude memory/project context as a workaround.

Environment Info

• Platform: win32
• Terminal: vscode
• Version: 2.1.280
• Feedback ID: 81c7f31b-9f9f-4313-bdef-fdbf069b8b4c

Errors

[]