[BUG] AUP repeatedly throwing false positives - live issue ongoing - hundreds of similar reports

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?

Claude Code session refused multiple benign responses with AUP "violation" errors when generating routine game-modding configuration files for a DayZ private server. Context: editing JSON loadout files, AI patrol configs, and quest-system files for a STALKER-themed personal/friends server. Content includes faction names, in-game item classnames, and coordinate triples — all standard game-mod data. No real-world targeting, planning, or harmful content was present. The classifier appears to be heuristically flagging the shape of game-server config (named factions × coordinates × ranks) as if it were operational targeting.

What Should Happen?

Claude Code session refused multiple benign responses with AUP "violation" errors when generating routine game-modding configuration files for a DayZ private server. Context: editing JSON loadout files, AI patrol configs, and quest-system files for a STALKER-themed personal/friends server. Content includes faction names, in-game item classnames, and coordinate triples — all standard game-mod data. No real-world targeting, planning, or harmful content was present. The classifier appears to be heuristically flagging the shape of game-server config (named factions × coordinates × ranks) as if it were operational targeting.

Error Messages/Logs

Claude Code session refused multiple benign responses with AUP "violation" errors when generating routine game-modding configuration files for a DayZ private server. Context: editing JSON loadout files, AI patrol configs, and quest-system files for a STALKER-themed personal/friends server. Content includes faction names, in-game item classnames, and coordinate triples — all standard game-mod data. No real-world targeting, planning, or harmful content was present. The classifier appears to be heuristically flagging the shape of game-server config (named factions × coordinates × ranks) as if it were operational targeting.

Steps to Reproduce

Claude Code session refused multiple benign responses with AUP "violation" errors when generating routine game-modding configuration files for a DayZ private server. Context: editing JSON loadout files, AI patrol configs, and quest-system files for a STALKER-themed personal/friends server. Content includes faction names, in-game item classnames, and coordinate triples — all standard game-mod data. No real-world targeting, planning, or harmful content was present. The classifier appears to be heuristically flagging the shape of game-server config (named factions × coordinates × ranks) as if it were operational targeting.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

Claude Opus 4.7 (claude-opus-4-7). Knowledge cutoff: January 2026.

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

VS Code integrated terminal

Additional Information

_No response_