[BUG] Fable 5 safety classifier auto-switches to Opus mid-session during authorized, in-scope security testing

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?

The Fable→Opus auto-switch also fires on non-offensive turns inside an authorized engagement — reading a target's own JS bundles, analyzing
▎ in-scope API responses, and drafting the disclosure report itself. None of that is exploit/malware/tooling construction. For verified CVP participants doing contracted,
▎ in-scope testing

What Should Happen?

please honor CVP status to keep the selected model in-session rather than switching per message — the mid-session flips break continuity on long engagements.

Error Messages/Logs

[]

Steps to Reproduce

e

1. Begin an authorized engagement on Fable 5.
2. On turns involving crafted requests — and sometimes benign analysis — the message is flagged.
3. The model silently switches to Opus 4.8 mid-task. Re-selecting Fable doesn't stick; the

next flagged turn flips it again.

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.70

Platform

Anthropic API

Operating System

Other Linux

Terminal/Shell

Terminal.app (macOS)

Feedback ID: 5daa4214-de6f-422c-8377-024940d8395b

Additional Information

_No response_