Execution Risk and System Control

What does Default-Block mean?

Detailed answer

Key answer: Default-Block means that high-consequence execution should not proceed when the required conditions for judgment and governance are absent.

Core explanation

Default-Block means that high-consequence execution should not proceed when the required conditions for judgment and governance are absent.

It is one of LERA’s most important principles.

Default-Block does not mean that every system action is permanently denied.

It means that execution is not automatically permitted merely because an AI system can generate a command, recommendation, or plan.

For a high-consequence action to proceed, the system must have sufficient grounds to allow it.

Relevant conditions may include:

  • legitimate authority;
  • anchored responsibility;
  • applicable and valid rules;
  • acceptable consequence level;
  • sufficient reliability;
  • adequate information;
  • acceptable uncertainty;
  • a valid execution path.

If these conditions are not satisfied, the appropriate result is not automatic execution.

It is:

Block

or

Escalate

This reverses the dangerous assumption that machine capability should proceed unless explicitly stopped.

In LERA, high-consequence execution must earn permission.

For example, an AI-controlled energy system may propose discharging a large battery asset because the electricity price is high.

If the action satisfies operating rules, safety limits, reserve requirements, authority conditions, and responsibility requirements, it may be allowed.

If the proposed discharge would reduce emergency reserve below an acceptable level, conflict with thermal safety rules, or exceed authorization, the action should be blocked.

If the information is incomplete or the consequence is unusually high, it should be escalated.

Default-Block therefore creates a disciplined execution posture:

when the system cannot establish that high-consequence execution is governable, it does not proceed by default.

This principle is especially important in systems where errors are irreversible, difficult to repair, or capable of affecting people, assets, infrastructure, or institutions.