I doubt that this is true except maybe for the top journals. Mid and low tier journals cater to scientists whose main incentive is to publish no matter how while moderately optimizing for impact factor (i.e. readers and citations). This lower quality market is huge. The fact that even top tier publishers have created low-ranking journals that address this market segment using APC-based open-access models shows the alignment between publisher and author interests will not necessarily lead to increasing quality, rather the opposite.

Does anyone actually read articles from those low tier journals? Many of those articles are illegible fluff pieces.

That top tier publishers create new low-tier journals for this market shows that they are very well aware of these incentives and risks. They are not flooding their top journals with low quality OA "pay to publish" articles, which was the argument from OP.