That's a bizarre and frustrating situation! It defies normal GFCI operation. Here's a breakdown of the possibilities and why each could contribute to such odd behavior

Understanding Normal GFCI Operation
Normal Operation

A GFCI constantly monitors the current flow between the hot (line) and neutral (load) conductors. If there's a difference of 5mA (milliAmperes) or more between these currents, it assumes current is leaking to ground (a potential shock hazard) and trips, interrupting the circuit. The "reset" button re-establishes the connection when the imbalance is gone, and the "test" button creates a small imbalance to simulate a fault.

Line Side Power Up

When power is applied to the Line side of a GFCI, it should generally self-test and, if all is well, allow current to flow to the receptacle outlets without needing to press the Test or Reset buttons.

No Load Connection

Without anything wired to the Load terminals, the GFCI should essentially be acting as a very basic, self-monitoring receptacle. There's nothing downstream to cause a ground fault unless there's a fault within the GFCI itself.

Possible Explanations (Ordered by Likelihood)


Faulty GFCI (Most Likely)

This is, by far, the most probable cause. GFCIs contain electronic circuitry. Even new ones can be defective. The internal monitoring and switching components could be malfunctioning.

Specifically

The sensing circuitry might be requiring a simulated fault (pressing the Test button) to "prime" it into an active state, and requiring the Reset to be pressed as some sort of abnormal "kill switch." This is just a theory, but it suggests deep-seated malfunction.

Manufacturing Defect

A short or mis-wiring inside the GFCI during manufacturing could cause this reverse behavior. Sometimes components are placed backwards or connections aren't properly made.

Damage During Shipping/Handling

While less likely, rough handling could damage the sensitive electronics inside.2.

Residual Capacitance/Inductance (Less Likely, But Possible)

Even with nothing wired to the load terminals, there's a tiny amount of capacitance and inductance in the wires themselves and in the GFCI's internal components. In extremely rare circumstances, this could cause a phantom voltage that the GFCI's sensitive electronics might be misinterpreting. However, this is very improbable.3.

"Ghost Voltage" Issues (Unlikely in This Scenario)

Under specific conditions, long, parallel wire runs, or proximity to other energized circuits can induce very small voltages in a disconnected wire. A high-impedance voltmeter might detect a few volts (sometimes called "phantom" or "stray" voltage). GFCIs are designed to ignore such tiny voltages. Because you pressed the test and reset buttons, this is highly unlikely.4.

Mis-wired Line/Load (You said it was correctly wired, but double-check)

The most common GFCI issue is accidentally wiring the Line (incoming power) to the Load terminals. However, you clearly stated you only connected to the Line side and did not wire to the load terminals. Also, if you did, you would probably trip the GFCI.5.

Extremely Unusual Grounding Problem (Very Unlikely)

If the ground wire itself had a high resistance, it could potentially cause a slight voltage differential with the neutral, creating an unusual feedback loop. But this would only be possible with a seriously bad ground connection.

Why Replacing It Was the Right Decision

Given the abnormal behavior, replacing the GFCI was the correct course of action. GFCIs are safety devices, and any deviation from normal operation should be treated with extreme caution.

In summary, the most logical explanation is a manufacturing defect or internal electronic failure within the GFCI itself. Your experience is unusual, but faulty GFCIs do happen. Flag for review