It means the true radiocarbon age of the sample falls within a statistical range centered on 10,675, with 95 percent probability of being within ±95 years of that central value.

But radiocarbon years are not calendar years.

The atmospheric production of carbon-14 has varied over time due to changes in solar activity, geomagnetic field strength, and other factors.

The calibration curves that convert radiocarbon dates to calendar years correct for this variation, but the calibration is itself imprecise, and the imprecision is especially significant at certain periods in the past where the calibration curve is relatively flat — where large changes in calendar years correspond to small changes in radiocarbon years, producing a situation where a specific radiocarbon date might correspond to a wide range of actual calendar dates.

The terminal Pleistocene — roughly 12,000 to 13,000 calibrated years ago — is a period where calibration uncertainty is real and where the translation of radiocarbon dates to calendar dates should be done with the specific calibration curve used and with an explicit acknowledgment of the resulting uncertainty range.