There is probably a linear range of phosphate measurement, but when the phosphate concentration gets too high, the detection reagent becomes depleted so that the standard curve flattens out. If your samples have too much phosphate to be within the linear range, dilute them before measurement.

Actually, in a strict sense nothing follows the Beer-Lambert law as it was formulated, since absorbance depends non-linearly from concentration even in the absence of chemical interactions.

Article Beer's Law‐Why Integrated Absorbance Depends Linearly on Concentration

However, luckily, in spectrophotometry quite often a linear range is found. This does not mean, however, that you always have to draw a straight line through your data - use a non-linear fit! As long as you do not extend it into a region where there is no detectable absorbance change any more (or the same absorbance value for different concetrations), you still should be able to obtain the concentration...