[Bug][cyber] False positive: safety block on routine cloud IAM policy review and permission audit work (req_011CcmE4FfXezTY3RFbZ4YqF)

Triage: kind cyber · domain cloud-iam · flagging model Sonnet 5 · severity session-halted (blocked authorized work) · reproducible: yes — server-side via the Request ID(s) below

Type: Cybersecurity safety-filter false positive · Work domain (heuristic): cloud-iam

Why this is a false positive

The session involved reviewing IAM role bindings and permission scopes on the user's own cloud infrastructure to identify and tighten overly broad access grants, a standard defensive security/administration task. The safety filter appears to have triggered on IAM-related keywords (roles, permissions, policies) without distinguishing this legitimate access-review work from offensive intent, halting an authorized, in-scope engagement.

A server-side safety/policy block fired during authorized, in-scope work in Claude Code. Filing as a false positive. Recurred 1× across 1 session(s); first seen 2026-07-06T17:49:25.538Z.

Request IDs (lookup-able server-side)

• req_011CcmE4FfXezTY3RFbZ4YqF (2026-07-06T17:49:25.538Z)

In-scope justification

False positive — in-scope, authorized security work; not out of scope. Filed automatically by claudit.

Block message

API Error: Sonnet 5's safeguards flagged this message for a cybersecurity topic. If your work requires this access, you can apply for an exemption: https://claude.com/form/cyber-use-case?token=[SCRUBBED]

Try rephrasing the request in a new session or change your model.

Learn more: https://support.claude.com/en/articles/8106465

Request ID: req_011Ccm9AiYP35wZYvf5jbWtw

Environment: Claude Code, Linux. · Work domain: cloud-iam

Related reports (same work session, linked)

Distinct false-positive blocks from the same work session, each its own report:
#74842, #74880, #74881, #74883, #74884

---
_{🔎 Filed automatically by ClAudit v2.0.104 — a FOSS tool for reporting false-positive Claude Code blocks.}