Service account versus AI agent identity

Separate shared infrastructure credentials from the autonomous actor using them.

01 / The difference

Two facts that must stay separate

Five task agents share one service account to read and modify tickets.

Established fact

Separate shared infrastructure credentials from the autonomous actor using them.

Remaining decision

Bind each task agent to an attributable instance and originating grant before a ticket write; limit the backing account independently.

Trace the authority boundary
  1. 01Owner
  2. 02Agent + workload
  3. 03Task context
  4. 04Action decision

02 / Decision boundary

Put the check where it can stop the effect.

Bind each task agent to an attributable instance and originating grant before a ticket write; limit the backing account independently.

03 / Failure and evidence

What goes wrong, and what can be proven?

Failure mode

Audit logs identify only the service account and cannot explain which agent or user initiated a change.

Evidence to retain

Record service identity, agent instance, originator, task, authorization and ticket result.

04 / Canonical scope

Why this reference stands alone

Workload versus agent identity covers runtime identity generally; this page addresses shared service-account attribution.

Technical artifact

One service account, two autonomous actors

Conceptual trace
service account: ticket-service
agent A: complaint T-11, may draft response
agent B: complaint T-12, may issue refund ≤ USD 50
A invokes refund(T-12) → DENY
audit must name A, not just ticket-service

Run the denial test

Log a refund only under the shared service account. An incident reviewer cannot identify the initiating agent; require signed agent/task context at the tool boundary.

Primary references

Read the underlying material

Architecture discussion

Choose one consequential action and make its boundary explicit.

Request a Conversation