> the algo obviously suggests it back to her

Pro-suicide content has been a thing since the days of USENET. It's been a (lesser) problem on social media since the start. Is it really just that one case got over the hysteria threshold?

But yes, if the algorithm is suggesting pro-suicide content then the developers are morally, if not legally, liable for that and should expect some consequences. I note that one of the few taboos maintained by the otherwise grossly irresponsible UK media is not reporting on suicide (because this is known to be a trigger). You might see "famous musician died suddenly at a young age" and have to connect the dots yourself.