Automatic signals with emergency replacement won't be plated as autos, so that a driver can't pass on his/her own authority as the signaller may be holding the signal at danger for a good reason.
The term 'automatic signal' has two distinct but related meanings, which can be a little confusing!
In signal engineering 'automatics' are signals used to divide up sections of unidirectional running line between junctions into distinct shorter blocks, so as to provide headway for traffic closely following in the same direction. Because there are no junctions in each block, no signalled conflicting movements possible in the other direction, and signaller workload concerns favour automatic re-clearance after the passage of each train, the logic in the control system to show yellow is remarkably simple, incorporating only track circuit clear to next signal overlap and proof the next signal is alight. To get better than yellow the aspect of the next signal also needs to be known, and an emergency replacement facility by the signaller can also be included, but this was by no means fitted to every example traditionally, often typically only one in five. Because of this simplicity, in old relay installations the entire control logic could be incorporated in the lineside cabinets and cables, with no centralised interlocking necessary and only very limited distributed remote control and indication facilities. Almost all such plain line signals were plated as 'automatic' traditionally, but a small number were not so classified because of particular location specific risks.
In operational terms an automatic plate has nothing to do with the control architecture, but is a clear indication to the driver that passing that signal at danger is allowed under certain circumstances without specific instructions from the signaller. The idea of the 'auto' plate was to allow traffic to keep moving if a signal is at danger and the driver couldn't get in touch with the signaller to obtain authority to pass it. On unidirectional lines there's little risk involved as the train passing the red must proceed able to stop at any obstruction, and traffic can't approach from the other direction except under special procedures where the signaller must ensure only one train is in the entire single line section at any time anyway.
Today, drivers usually talk to signallers over cab radio and if that fails the lineside signal telephones can provide back-up communications. There's little chance of no conversation taking place now, so the marginal risk of driver self-authorised movements, and their often slower speed can be avoided, hence auto plates are often not provided in new projects. Modern signal control technology allows every headway signal to be provided with an emergency replacement facility.
The train detection systems that control headway signals are also evolving, with the traditional track circuit giving way to axle counters in many areas. With direct connections to the exposed running rails and critical 'knife edge' adjustment of operating voltage and current, track circuits can be fairly prone to 'right side' failures related to weather or cargo spillage, and remain a significant reason why signals show red unintentionally.
Axle counters are much less prone to failures of this sort and should further reduce the need for trains to pass danger signals.