Subtle adjustment, I think the real cause is scaling human relationships. IT enables this scaling. So if interpersonal trust scales sub-linearly with number of connections, it's not really IT corroding trust, it's what IT enables.
It's ... complicated. More than I can get into here, or eveen than I've really worrked out.
Short version is that in the absence of advanced comms, you have no option but to extend trust. I'm thinking here where inteervals might have been months, eveen years. (Historically, sea passage was not possible from November through May.)
Instead you had trust-establishing institutions, a role usually filled by religion, in virtually every pre-industrial civilisation.
Comms, from the scroll and telegraph forward, obviate this. Wearables, CCTV, desktop tracking, cellphone monitoring, all substitute surveillance for trust.
Trust isn't a binary, it can be extended to differing degrees. In absence of advanced comms, you have to extend trust, but how far? That comes with an opportunity cost -- I wouldn't loan money to someone I'd only communicated by 8 week delayed handwritten letters, but I (possibly) would do that to someone I'd talked to on Skype for an equivalent amount of time.
They also come with significant opportunities for fake signaling.
Gaining trust means you signal properly to the people you are attempting to influence. With records you now move mechanisms with which your signal that trust.
False signalling is an issue, but that's hardly novel. Mimicry, imitation, camoflage, and other forms of disguise and deception are found in nature.
There's the matter of taking what had been very strongly analogue, that is, analogous, forms of recordmaking and turning them into mediated forms: static images, audio, and video. But that is merely a regression to the status quo ante of preindustrial times. The magic of photography, tape, and film were that these, more or less, faithfully recorded what was before the lens or the microphone. We're returning to a period that bears more in common with the age of human testimony, at least in terms of its level of mediation. That is: you have to trust in the testifier. The distinction in the quantity (and detail) of the images presented is not as in earlier times.
It's the fact that communications cycles and volumes are so much greater though that reduces the reliance on trust, in ways I'm starting to suspect are more substantive than is generally appreciated.
There are several histories of corporate communications, where this trends appears, to an extent. James R. Beniger's The Control Revolution (1986 or thereabouts) and JoAnne Yates, Control Through Communciation (2005) are two of which I'm aware. The role of trust isn't central to either, but it's touched on.
Fake signalling is what's always been present in communications. Deception, bluff, camoflage, and distraaction are all highlydeeveloped even aamongst animals. It is scale and rate especially which are changed now, and the ability to entirely bypass trust or trust-development.
That's not a solid conviction, but it's how I'm leaning.