There are plenty of other online platforms where you can use the unmodified model without siphoning your data to China. The model itself is just an offline blob and doesn’t need to be modified to make a “more secure” and “privacy friendly” version like the article says it does, because the model is not tasked with collecting and sharing your data. The author doesn’t seem to be aware of that.
The platform you’re using is indeed an online platform with its own data practices, but that’s common sense and has nothing to do with this model being Chinese. It’s up to the user to decide wether or not to use online AI services at all and which ones (not) to trust. The model itself isn’t doing anything with your data.