Authority is usually understood through people.
A CEO has authority because of their position. A Board has authority because an organisation formally grants it powers. A Regulator has authority through law. A Manager has authority over particular decisions because those decisions have been delegated to them.
But there is another form of authority that matters increasingly in a world organised through software and infrastructure: the authority to make something actually happen.
This is execution authority.
Execution authority is the capacity to determine, initiate, constrain or enforce consequential outcomes within a governed domain, without requiring human authorisation at the point of execution.
The distinction between authority and execution authority matters.
A person can possess formal authority without possessing the practical ability to produce an outcome. Conversely, a system may possess no legal title, organisational position or recognised institutional authority while still determining what can and cannot happen.
Consider a payment.
A bank Executive may formally have authority over the institution. But when a customer attempts a transaction, the Executive does not personally decide whether it proceeds. Systems authenticate identity, assess conditions, apply limits, detect potential fraud and ultimately allow or prevent the transaction.
The consequential moment is execution.
The system does not need to possess consciousness, intention or legal personhood for this to matter. Nor does it need complete autonomy. Humans may have designed the rules, selected the technology and established the constraints.
The important question is simpler:
What determines the outcome at the point at which the outcome becomes real?
This distinguishes execution authority from ordinary automation.
A calculator performs a task. A printer executes a command. Neither necessarily exercises execution authority.
Execution becomes authority when the capacity to act can determine, constrain or enforce consequential outcomes within a domain: whether someone receives access, whether money moves, whether an application progresses, whether a transaction is permitted, whether infrastructure operates, or whether an institutional action occurs.
Execution authority therefore exists on a spectrum.
It can remain predominantly human. It can be distributed between humans and systems. Or substantial parts of it can become embedded within software and infrastructure.
This is why artificial intelligence matters, but also why execution authority cannot be reduced to AI.
Software systems were exercising forms of execution authority long before generative AI. Payment networks, identity systems, trading systems, cybersecurity infrastructure, platforms and automated organisational processes already determined what could happen without requiring a human decision at every point of execution.
AI changes the scale and character of this phenomenon because systems can increasingly interpret information, infer, generate responses and participate in decisions before initiating action.
The result is a growing separation between formal authority and executable authority.
An organisation chart tells us who is authorised to decide.
It does not necessarily tell us where the capacity to produce the outcome resides.
Governance must therefore ask a new set of questions: Where does execution authority reside? Who allocated it? What constrains it? Who has custody of the infrastructure carrying it? Who can intervene, reverse or override its actions?
Execution authority gives us a way to locate authority where it actually operates.
And increasingly, that means looking beyond the person and into the system.
Researcher: Mamatsabu Maphike