Because the academic standard for research, and the original motivation for tenure, allows for researchers to criticize anyone, including their own institution, as long as it's done and presented as the proper product of research. In other words, if a researcher is doing their research properly, it shouldn't matter if they're attacking their own institution or even colleagues, because it passes scientific muster and the quality of the research is what justifies it.
When companies like Google or IBM employee researchers, they're positioning themselves as supporting that model of inquiry, and the benefit for them is the credibility that comes from participation in scientific inquiry. Gebru attacking Google's AI practices actually lends Google credibility for disinterestedly supporting basic research and taking its results seriously.
A company that employs an ethicist and takes that ethicist's output seriously even when it challenges the company is a company that genuinely cares about the academic standard in that field. A company that hires an ethicist and then fires them when they point out the company's wrongdoing is sounding an alarm about their own unethical conduct.
The part I struggle to understand here is how Google could be so stupid. Who cares, from a commercial perspective, if an academic they employ drags them? Keeping Gebru on payroll was straightforwardly purchased ethical credibility; firing her is a PR disaster and an open statement that their ethics AI group is a propaganda unit rather than a genuine research entity. It would have been trivial, even beneficial, to air the academic dispute. They gained nothing by retracting her paper and then firing her.