As I said, I have already been doing this for many months on a Full Spectrum 90W CO2 laser at my Makerspace. I had to research my method in advance and prove it to the board. I not only did that, I got the sign off of Full Spectrum’s tech support.
The issue is indeed the reflectivity, and is the reason you can’t use a diode laser for this.
The beam is scattered back to a degree. In a diode laser, that is right back at the emitter. In a CO2 laser, the laser is at the end of a chain of mirrors and is very far away from the focal point. There is minor risk to the mirror in the track head, but a good quality mirror should be fine.
I can also prep the surface of the metal by giving it a roughened surface, which diffuses what little reflections there are.
Choosing the right lens is also important, you want one with the smallest focal height you can manage. Again for slight diffusion purposes.
I am not using the laser to do anything to the metal. I am running at 15% power, as I’m using the laser to ablate a coating off of metal. I then use chemical or electrical methods to etch the metal, and then I remove the coating.
I understand that this is a risk of a certain level with the Glowforge, but I believe that based on what I know about the design it should be fine. I will, however, confer with support.
EDIT: Copper is VERY reflective in the IR Spectrum. So is Silver, but Copper is worse. You need a fiber laser to do any direct metal removal, either cutting or ‘etching’. What a lot of people refer to as ‘etching’ is using a coating like Cermark, which the laser then bakes into a permananet ceramic coating on the metal.
Those designs in my pieces are from .5 to 1 mm deep, and were removed by chemicals, not the laser.