That's not true. From the article, type 2 is more familial than type 1, but not all of that association is genetic.

As a more concrete demonstration, the type 1 genetic risk score (GRS) has good predictability of the risk of someone getting type 1 diabetes. We have linked certain genetic variants to increased or decreased risk of getting type 1 diabetes (and it's mostly in the HLA complex on chromosome 6 that significantly influences the immune system). The AUC (area under curve) of the score's ROC curve is 0.87, which is good. We use the type 1 GRS for patients incoming with type 1-like symptoms to separate out those likely to have rare genetic conditions instead, alongside antibody testing, and it works very well.

The type 2 GRS is very weak in comparison. We haven't found much link between genetics and type 2 risk that we can use to predict the risk. The AUC of the type 2 GRS is only 0.63 in the very best studies, which is a poor predictor.

Having said that, type 2 risk varies quite considerably with race, with South Asians being more susceptible to type 2 diabetes than much of the rest of the world.

Note, an AUC of 0.5 indicates no predictive value whatsoever, and an AUC of 1.0 indicates perfect prediction.