Executive Authority Brief · VOL 2026.02

Delegated Authority

Where Autonomy Begins

By Hadi Hendrawan — Advising CEOs on AI Risk, Authority & Accountability · March 9, 2026

Download PDF Open PDF

Implicit Delegation

Example

The system operates with broad objectives but undefined execution limits.

Authority

Assumed.

Consequences

  • Ambiguous boundaries.
  • Reactive oversight.
  • Post-incident clarification.

Control

  • Informal supervision.
  • Case-by-case intervention.

Explicit Delegation

Example

The system operates within predefined financial and operational thresholds.

Authority

Defined in advance.

Consequences

  • Predictable execution scope.
  • Clear accountability locus.
  • Scalable oversight.

Control

  • Written mandate.
  • Quantified limits.
  • Defined override rights.

What Changes at Scale

Undefined delegation expands silently.

Defined delegation contains exposure.

Design determines defensibility.

ACTION: Before Delegating Authority

  • What decision rights are transferred?
  • What thresholds define execution?
  • Who retains override power?
  • How is delegation documented? Autonomy scales. Authority contains.

Hadi Hendrawan

Advising CEOs on AI Risk, Authority & Accountability

March 2026

SUPPLEMENT: Authority in Personal AI vs

AI Agents in Group Conversations

HIGHLIGHTS

AI can remain a personal tool.

Or it can become an organisational actor.

The difference is where authority sits.

When AI agents sit inside group conversations, authority may be exercised WITHOUT explicit delegation.

No explicit delegation. Yet actions occur.

Authority becomes AMBIENT:

Responding to instructions that emerge from conversations rather than a defined authority.

Personal AI AI Agents in Group

Conversations

Example: ChatGPT.

Environment

Private interaction between an individual and the system.

Function

  • Drafting.
  • Research.
  • Reasoning support. Example: OpenClaw.

Environment

Organisational communication channels/groups: Whatsapp, Telegram, Microsoft Teams, Google Chat.

Function

  • Information retrieval.
  • Task execution.
  • Workflow triggers.

Authority

  • None.
  • The system produces suggestions.
  • The individual decides.

Consequences

  • No organisational delegation.
  • No operational execution.
  • Contained risk.
  • AI supports thinking.
  • Humans retain authority.

Authority

  • Participants may include internal staff, external partners, clients with different levels.
  • Operational presence inside live conversations.
  • Instructions may arise from any participant.

Consequences

  • AI becomes part of organisational operations.
  • Authority may shift from formal instruction to conversational cues.

Governance Exposure

AI agents embedded in group conversations introduce new governance risks.

Authority Drift

  • A request appears in discussion.
  • No formal command, the agent acts.
  • Responsibility becomes unclear.

Role Confusion

  • Participants hold different authority levels.
  • The agent may treat all instructions equally.
  • Authority hierarchy disappears.

Information Spillover

  • Group conversations often mix:
  • Internal information.
  • External participants.
  • Operational discussion.
  • An automated response may disclose information beyond its intended boundary.

ACTION: Before Adding AI Agents to Groups

  • Who is allowed to direct the agent?
  • What authority does the agent actually hold?
  • What information must the agent never disclose?
  • Who is accountable if the agent acts incorrectly?
  • How can leadership immediately stop the agent if necessary? Autonomy scales. Authority contains.

Hadi Hendrawan

Advising CEOs on AI Risk, Authority & Accountability

March 2026