Can we expect similar issues such as spectre and meltdown that intel experienced with speculative execution.. but, in the form of prompt injection/poisoning?

Ok, I'll bite: no, considering these are very different domains and you don't get system access by getting the wrong speculative branch for your next text token, you just get a slightly different (but probably still related enough) text.