It may not be as troubling as at first look, if we consider that the current 370 CO2 is only 22 percent higher that the highest 304 from Vostok ( as I recall from the table of data) (close to this on graph).
We don't know the accuracy of CO2 and Temperature data from 300,000 years ago. I suspect +/- 22 percent accuracy of the data would be very good. The ~304 years ago may have actually been ~370.
In fact, a couple of the web references state that the CO2 concentrations lag the temperature data by 1000 to 4000 years. This means that we need to learn much more.
But it all is very interesting.
CO2 concentrations are not calculated from proxies; they are directly measured from bubbles trapped in the ice, and therefore are probably quite accurate. The temperature is determined from stable oxygen isotope ratios and is more uncertain.