Here in Canada, it has lately gotten hard to see the phrase "government scientist" without seeing the word "muzzled", somewhere within a 5-word radius. The press, and advocacy groups love the expression, and use it at every opportunity. I'll leave it to you to dig up the particular conflicts, using those key words.
The conflict, as I see it, is that between two very different professional world-views that do not abide under the same roof very easily.
The world of policy is one that demands uniformity and consistency. Citizens, other government agencies, other levels of government, businesses, investors, advocacy groups, other nations and international bodies, all want to know "What is the policy direction of your agency/government?". That direction might be completely pragmatic, partisan, or not. Whatever the case, comms people strive to present a coherent and consistent policy stance to all stakeholders. Diverging from prepared media lines, and timetables of spooling out information, works against this goal.
In another parallel universe, scientists are trained to value anomalies, and instantaneous discussion. It is the surprising and unexpected, after all, that move science forward and leads to better theories. The possibility that a paper or result or observation announced on a Monday, might conflict with another paper, result, or observation, 3 Fridays later, is not cause for concern, but merely cause for reflection about what ought to be tweaked in the theory. What is of value is not uniformity of findings, but the pattern observed across the preponderance of accumulating evidence. That's how we're trained and how we think.
So comms and policy people want any and all work by government scientists to be more aligned and uniform than it is or even can be, and don't want it to gum up whatever policy announcements are planned. That's not wrong, it's just how they work; I don't blame tigers for not eating more salad. Scientists, on the other hand, want the freedom to engage in collegial interaction and present their work for discussion at the earliest possible opportunity. What I might know about offshore fish stocks, or glacial melting, or livestock pathogens, or food safety, provides benefit by being presented and discussed in as timely a manner as possible, warts and all. That timetable may not align with the political timetable, even if the findings don't conflict with the policies (although they might).
The conflict arises because neither can afford to ignore OR completely yield to the other. Policy that changes with the latest data, or that conflicts with science from its own staff, stops being an identifiable direction that external stakeholders can rely on with confidence. Science that proceeds on its own timetable without connection to the public interest, and its expression in policy, becomes merely academic (not a sin, just better-placed in another context). The two need to get along better, and that requires compromise on everyone's part.
I mention this because there IS a responsibility to represent one's institution in a manner that conveys coherence of vision and direction, and avoids confusion. On the one hand, it should be possible to say "Personally, I think the policies need a tiny bit of tweaking to do as good a job as we hope for, but all in all this is the direction we're heading in." On the other hand, media likes a juicy story and conflicts always provide juice, as your own experience indicates. Qualified agreement seems like a no-brainer, but it can be too tempting for some to focus on the qualifications, and ignore the agreement.
You gots to be careful.