Hmm, those aren’t half-bad ideas actually. Perhaps with some creative use of optical sensors that could be done.
The print getting knocked off the bed would be easier, just set up optical gateways (like the “bing-bong” pass through detectors in convenience stores) around the sides of the bed. After a print starts, if anything trips that, PAUSE!
The print attached to the head would be a little trickier to do since there are so many ways a piece of gum can get stuck to something I wonder though, that’s got to somehow change the resistance or heating capacity of the nozzle somewhat. I’m just not sure if there’s enough capability in firmware to measure that. I know Marlin firmware is getting really good at detecting thermal runaway, but that’s a much more pronounced effect. A slowly building wad of filament is much more subtle.