Pretty sure the issues with direct injection, namely the gradual occlusion of the intakes, was nailed down to certain users overly-frequently changing oil and using improper (ie: too volatile) lubricants for the job. In a nutshell, a petrol engine acts as a sort of vacuum pump on the crankcase through the PCV, and the volatile components of motor oil eventually find themselves in the intake system. In engines with fuel injection into the intake air itself, the fuel would act as a solvent and keep deposits to manageable levels with the odd throttle body cleaning. However, with direct injection, there is no natural 'cleaning' for the intake.
Implementations were tested by the manufacturers based on manufacturers' spec'ed lubes, and manufacturer spec'ed OCI's. No problems were detected. I think the manufacturers under-estimated the non-adherence of the operator community to both the recommended OCI's (more frequent isn't necessarily 'better', and can actually be, in such case, far worse for an engine), and even to the recommended oils (even dealers are often guilty of using the cheapest bulk 5W-30 dino oils they can find when they service very expensive cars!).
A couple years ago, I did a meta-analysis in a post here, where I took a thread of car owners complaining of the DI intake occlusion problem in a car-specific online enthusiast forum. And examined their stated OCI's. All of them were dramatically in excess of manufacturer's recommendations, for no reason other than the irrational belief that changing more oil more often would prolong engine life.
GM has pushed Dexos, API tightened down the volatility spec, and most vendors have engaged in campaigns to educate their operator communities that oil changes in excess of the specifications and/or OLM are not required and may be of significant detriment. I believe there has also been additional activity in terms of auditing dealers for compliance with specifications.