What is AGI Safety?

AGI Safety is not simply about making AI “safer.”
At its core, it asks how a system with near-general intelligence can remain controllable in its judgment, execution, and consequences.

This is not only an algorithmic problem, and not only an ethical one.
Once AGI becomes connected to real-world devices, energy systems, industrial operations, and critical infrastructure, safety moves from the digital domain into the physical domain.

This is the context in which LERA is proposed.
Rather than asking only what AI should do, LERA asks under what conditions AI may execute, and who has the authority to permit that execution.