Natural reproduction already has a lot of randomness and failure built into it. It's not like egg/sperm selection is a perfect quality-control system. IVF has also been around for decades, and while it definitely needs monitoring, we haven't seen some obvious collapse in health from it
I'd be very curious how they even define "safe enough" for this. With most therapies, the risk is mostly to the patient. Here the risk could be passed on to a future person who never consented to the experiment.