Shouldn't the same principle apply to mentioned companies and law enforcement agencies as well then?
That being said, I think you're really misusing the "innocent until proven guilty" idea. If a complete stranger appeared on your doorstep wielding a steak knife and offered to help you prepare dinner in exchange for him staying the night on your couch, would you assume he was innocent, let him in, and then not keep a watchful eye on him? Most people would not, and this suggests that, all else being equal, most people would prefer to act in ways that don't put them at risk.
So I disagree that we should automatically assume that the intelligence community are all innocent. Nor do I think we should assume any of them are guilty with evidence. But I absolutely believe it's ok to watch them with scrutiny until we understand the implications of this data being used in investigations. Keeping a watchful eye on governments and companies encourages accountability. That's a good thing.
"At its worst, social media monitoring could create classes of 'pre-criminals' apprehended before they commit crimes if police and prosecutors are able to argue that social media postings forecast intent. This is the predictive business model to which Geofeedia CEO Phil Harris aspires."
And I think it's a decent concern. If somebody tries to build a profile of you out of your social media presence, they're most likely getting a pretty inaccurate (or biased) picture of you, which is especially concerning if they aren't aware of the limitations of these profiles - and judging by the history of things like polygraphs, this is likely the case. This potentially has consequences for your employment/background checks, or any time your character is called into question, in general.