Yes it is not correct. The crystal structure changes from FCC to BCT, and the symmetry (from a misorientation standpoint) changes from 432 to 422. However, any CSL boundary that lies along the C-axis of martensite could still fall under the FCC criterion along that axis.

Why is it done? Most likely because the C/A ratio in martensite is still close to 1 and provides a commonly used method to characterize grain boundaries. Whether there are any unique properties to the CSL (i.e. Sigma3) in martensite versus austenite, I have no answer.

Also note that the CSL criterion applies generally for cubic structures. The Sigma3 boundary in an FCC crystal is a 60° twist about the  axis, or 70.5° tilt about the axis. The grain boundary energies by CSL theory, although this is not a good correlation, for FCC and BCC do differ.

see this exemple  https://www.jstage.jst.go.jp/article/isijinternational/55/9/55_ISIJINT-2015-057/_html

Thanks very much for helpful information, dear V. Pertsevyi, B. Lin, H.E.B. Messauoud.

I have another question.

If a fraction of Sigma 3 boundary is high in martensite, an energy of the boundary is also low (similar to austenite) ?.

In my specimen, fcc single-phase transforms to the strain-induced martensite (bcc) single-phase, and all of bcc boundaries are Sigma3 and Sigma 11. It might come from K-S OR.

I wonder these boundaries are more stable than other high-angle boundary.