I mean, there are real projects. I know that the orca population near Vancouver is a focus of decoding mammal language.
They have recordings going back like 30 years and log books to connect the calls to orca behavior. Last I checked (a few years ago) they were pretty far with unsupervised learning on the audio data and we're going to tackle the (barely readable) logbooks next.
I doubt we are far enough to train ai for animal-human translation on more than a conceptual level, and I doubt that I would hear about it from dolphins for the first time.
I expect a widely covered story of translating dogs' barks (or cats) first, and not in a "Hello Human, Welcome back home, I missed you. Please give me food" way (which would be probably fake) but just "Friend! Joy. Hungry"
And I don't know how we could scientifcally differentiate a slur from a descriptive name on that conceptual level.
What I don't doubt is that dolphins have slurs for humans.
I am gonna take a wild guess (wo reading the original article), is this one of them studies where they throw a neural net at dolphins and see if it sticks.