Unfortunately, the mechanism to do this as you describe it doesn't exist for Access. You would have to craft your own method to do this. The normal way Access works follows the Windows security principle of user isolation; i.e. you have no right to interfere in the actions of another user. I know that technically, this isn't interference that you want, but rather a type of collaboration. However, to Windows, reaching out to alter the flow of operation of another user's process is expressly forbidden without setting up this collaboration first.
The normal data flow of an Access split database is that the front-end is the place where the workspace is physically located. The back-end is more or less passive here. It is merely a file being altered through the file system and the SMB protocol. There is nothing active in the back-end to send out such change notifications. EVERYTHING happens in the front-end. Therefore, it would become the responsibility of the front-end to send out such notices. But... there is again no easy way for that to happen unless the front-end files are programmed to cooperate in a way that is not native to Access. You can't just "reach out and touch" some other front-end file on some other machine. You have to program the front-ends to EXPECT this type of cooperation. When you are just sitting in front of the FE screen/form/whatever, no events are happening - and that means there is nothing to drive this sense of change. If you tried to open up a network socket, you still have the issue that whatever is monitoring the socket has to wait for it. (This is the "LISTENER" concept in network communications.) In the absence of driving events IN THE FRONT ENDS, you have no signaling mechanism because most of the time, ACCESS is in control of listening for things.
There is such a thing as a table data-change event - but it doesn't help because this event still executes in the context of the FE that changed the data. There is no active back-end code to watch for data changes.
If you have an active back-end like SQL Server or ORACLE or one of the other popular active SQL engines, there IS a possibility of having an active signal sent out, though again there is the issue of separation of tasks. Even if you have this hypothetical active back-end, the front-ends have to be set up to expect that it COULD happen. There is a LOT of programming required to do that.
Short answer: There is no intrinsic mechanism for this. You can't get there from here without writing a BUNCH of complex code to manage the "unsolicited" interaction. As experience as I am, this is something I wouldn't try without first stocking up on aspirin and coffee.