It's to do with the nuance between scientific and colloquial nomenclature, with simplifying established scientific knowledge for the layman, and even slips of the tongue.
First, a 'fact' in science is used more loosely than in strict epistemology: a fact is any established truth. In epistemology, it's anything which is strictly known to be true with 100% certainty; in science, it's known to very high (but not necessarily 100%) certainty.
The existence of atoms is a scientific fact: the evidence for their existence is so overwhelming that they've been proven to exist beyond all reasonable doubt. They aren't know to exist with 100% certainty, so they're not an epistemological fact, but they are a scientific fact.
The difference is similar to that between 'theory' in science and in the public eye; the former considers a theory to be a well-evidenced explanation, while the latter generally consider it to be just a guess.
Second, it is impractical to preface everything with "The evidence suggests", or "We think", or "We conjecture", or "We hypothesis", etc, and it also erodes the public's confidence in science. One need only to look to the USA to see how dangerous that erosion can be.
Third, some scientists (especially those talking outside their field) genuinely make mistakes. A biologist may say that the Big Bang theory "proves the universe began 13.5 billion years ago", despite that being false.