This is just a guess of mine, I don't know any data for it, so take it with a big grain of salt:

I can imagine that it is in each big greenhouse gas emitter's interest to state a lower estimate for how much GHG are released. Also many plants are reported of leaking gas.

So the data scientist use may not be the complete list of emissions.

Maybe.

They aren't really far off it's just that the models don't show every feature of the climate because they are models, not reality. But regarding the broad strokes the models are accurate and relevant.
Also CNN peddles fear and useless information. They thrive on sensationalism, not by informing people.