Not via simple legal declaration, no. Almost every system of land claim and titling in the settler-colonial world involved the additional retention of that information at a central office; usually, there was a fee involved in transfer of land, but the recording of that transfer was essential, because all sorts of other state functions (taxes!) depended on knowing who had possession. The systems used in British North America (and Quebec) were not entirely uniform--sometimes claims came first, with the prospect of overlap or contest, and then survey and title later; other times, it was the other way round. Regardless, once a title deed existed, it did not only exist in the hand of a landowner (if tenure decreed this was an actual owner and not a perpetual lessee). One might still take desirable land by fraud, coercion, or subterfuge, but they'd have to do more than just find a scrap of paper and seize it; the con would need to be bigger, as I note below. 'Claim-jumping,' a term we're likely more familiar with, is similar to what you suggest and might well fit the description. That involved more applying one's claim to a better parcel, because of the vagueness of the claims, more often than the theft of an actual claim. Nevertheless, it happened before the solemnization and registry (and thus issuance of a true deed), and was far more common with high-value lots than with homesteads. That's one reason why we associate it with the early days of mineral rushes before mine surveys, but in any settlement scheme for a new area (like western Virginia or NW Georgia) vague claim papers often would come before titles--sometimes many years before. (This system long prevailed in southern Africa as well, with similar chaos over who had the right to what, and where.)

By the 1700s in North America, it was also not just householders in such a system of titling either--it was also land companies, individual speculators, and other people who bought rights to blocks of land from individuals or from a colony. In order for the security of capital, the right of ownership needed to be seen as ultimately safe among the colonists, so maintaining deeds in county seats, and diagrams or deeds at the level of the colony (or state), was an important part of realizing the value of the land and being able to tax it. Edward Price's Dividing the Land: Early American Beginnings of our Private Property Mosaic (Chicago: University of Chicago Press, 1995) breaks out chapters roughly chronologically to talk about this evolution of titling, land-jobbing, and all of that. There's a little in Bill Hubbard Jr.'s American Boundaries (Chicago: University of Chicago Press, 2008) in terms of the superstructure and especially for the Rectangular Survey which flourishes after the era you ask about, but it may offer some illumination into registries. I don't have my copy of Hubbard with me, unfortunately.

In all of this, I have not really talked about two elements: the first, of course, is that the legal mechanisms might be subverted in one way or another. The title was proof that a landowner could present themselves, without having to dig into the registry system. If one did not have the means, or the knowledge, to force a legal challenge such a theft might stand if the thief could fool the registrars. The title would name the original awardee, and be amended if the land were sold (assuming, again, that it could be sold or transferred by the occupant--some land tenures were not freehold or quitrent, but leaseholds, etc), so in strict legal terms, simply taking the holder's copy of the title deed was insufficient to effect transfer--something else had to be going on scam-wise, and they had to have a way to effect the transfer before the wronged party could contest it. The details of stories you mention might give a clue: if someone presented themselves to county officials to transfer a deed they had in hand, and the lawful holder were (typically?) a widow or young heir, he (and yes, the gender of the two parties was quite relevant) might be able to hustle it. I don't know any specific US cases of this, as my work in the archives of Deeds and Survey offices has been South African. (There are cases there, usually between family members, although they usually settle out in front of the landdrost or magistrate.)

The second part is that a notional 'title,' which passed to a colony in order to give it the right to parcel out blocks and properties of land, was ascribed to Native Americans. This title was recognized only when forfeit, and rarely written down except to vacate it. When it was recognized via treaty or even (rarely) in a registry, its theft was far more likely on the grounds of disuse, misuse, or forfeiture for whatever reason colonists propounded. Again, this could vary dramatically, and John Weaver gets into that process comparatively across the Anglo world in his The Great Land Rush and the Making of the Modern World 1650-1900 (Kingston: Queen's/McMaster University Press, 2003). But that is also not quite connected to your question because, again, nobody stole a physical deed to do it--people in positions of power simply declared the right to land null and void, removed and/or dispossessed the occupants, and re-allocated it.

[quick edit for OP: Do you have particular stories in mind? I'm curious now--titling is a weird thing even for those who work with it, so the popular vision of titling has me wondering how such stories do portray it.]

[edit 2: Italics for Weaver; fixed a non-seq in para 4]