I won't say I trail however I am in the guard's van ready to apply the brakes. I certainly view connection of the ship with less regulated 'stations' ashore with equal concern. Remote control must mean an office as rigidly controlled as on the bridge and engine room of its ships, perhaps that will steer towards a hybrid or full autonomous solution.
The "don't do unless necessary" rule can be applied partially. Only allow communication between systems where that is necessary, do not have essential systems sharing a common bus with anything they need not be and extend that to communications.
Whether malice or error all electronic data whether received from a local source or a remote one may cause an undesirable system response. That even applies to the written word - Don Sixto on Conoco Europe was fond of recalling that there had been a galley fire and the office (London Brits) had asked for a report on the health of the staff (All Spanish). The ship resorted to the dictionary, "staff" being unrecognised, and translated this to lengths of wood as in stave. Clearly the staves above the galley deckhead were burned to a crisp and this is what was reported back.
Erroneous but not malicious GPS data controlling an AIS shore station caused Sperry's Visionmaster displays to crash when accepting AIS data, a systems failure which testing and FMEA had not discovered (an impossible date, leap year associated if I remember fully). Updating ECDIS on a Korean build LNG Newbuild caused the vessel to abruptly change course - incorrect procedure or perhaps an as yet undeveloped procedure).
I remain skeptical that these big solutions have corresponding big problems they are to solve. What is certain is that the do bring problems of their own. On the LNG vessel (mentioned above) on trials I was given a tutorial on Kongsberg's IAS and a long chat with their consultant. One of my bugbears with the way this sort of kit is applied to merchant ships is that, unlike the nuclear or public transport industries, it is applied without the same quality control - especially documentation. To me this is vital for a vessel expected to last 15 years as I contend that one may be faced with renewing any single failure prone silicon system at around 10 years. The silicon will be cheap enough but the engineering involved to marry the new to the ten year old will not be especially if the documentation of each and every one of its interfaces is not accurately detailed and recorded. I have opined here that the cost at this juncture might well mark the end of the economic life of the asset (not so much fatigue life as silicon-fatigue life). The consultant grinned and agreed "At ten years old this will not only be expensive to support, it will be fabulously expensive to support" - no prospect of him being out of work then!