There are several ways to hack such schemes.
In case of something trivial like this, create an alternate optical path for the sensor that reflects from a piece of the genuine foil top, with the “proper” ink. A little glue or sugru with small mirrors should do the job easily. No need to bother with modding cups.
If it is something more difficult, you can spoof the sensor itself, generate fake “genuine” response. A cheap microcontroller of Arduino class can handle a lot of such things.
Or you can break into the main firmware, hack the main CPU down to the bootloader, and inject your own modded code where the check always passes. It is usually all hinged on a single conditional jump that can be replaced with either an unconditional one or a NOP instruction.
The ultimate is throwing out the original controller and putting in a new one, keeping just the actuators and drivers and transplanting the brain. That could eventually be even sold as modkits. Given that a coffeemaker firmware is nothing exceedingly complex, that again falls into the domain of Arduino-class chips. And you can have all sorts of additional fun; if you splurge and put in a Raspi-class board, you can watch the coffee pot with an integrated camera, run the coffee cycle over the intranet, have all sorts of timing and scheduling, and (with either a prepositioned cup or a way of remote positioning, LEGO and some servos to the rescue) have an over-the-internet control so you can tell the machine what coffee you want before you get home, and be invited by a smell of a fresh cup.
…for added points, you can build it from scratch and have a coffee mill built in. Or even feed the machine with green beans and roast them per batch to an optimum degree. The disadvantages are a higher complexity than just a controller board replacement, and not showing a corporation that their silly schemes are trivial to bypass (and how would they learn).
