One of my early jobs was to build a custom C14 system using a "phoswich" detector. The use of two phosphors allows C14 radiation from the sample to be distinguished from exterior radiation. However it is imperfect and there is always a residual amount of exterior radiation that adds to a background level. The 40,000 year accuracy limit is due to unwanted contamination (eg fingerprints) and cosmic rays. The background can be estimated and subtracted, but the statistics of a null experiment show that the statistical variance can give false positives when stretched to the 40,000 year limit. Any C14 "detected" from old fossils are no doubt contamination or variances in the assumed background level. If you found uncontaminated fossils that had a CO2 level of the young earth 6,000 years, then that would prove your point, but as it is you can't support creationism from C14 measurement error in fossils. .