Is Chevallier-Polarski-Linder dark energy a mirage?

Recent observations from the Dark Energy Spectroscopic Instrument (DESI) raise doubts about the standard cosmological model, $Λ$CDM, suggesting a preference for an inherently dynamical dark energy component. The Chevallier-Polarski-Linder (CPL) parametrization -- a widely used two-parameter model for the dark energy equation of state -- displays marked early-time phantom behavior and a recent crossing of the phantom divide. These features suggest the convenience to check observationally the robustness of such evolution. To address this, we design two alternative families of two-parameter dark energy parametrizations which remain close to the original CPL but aim to soften its phantom character. Specifically, these models reproduce CPL-like behavior at low redshift but mitigate early phantom behavior through the use of smooth sigmoid transitions, yielding a more gradual evolution. By combining recent DESI data with constraints from the cosmic microwave background and Type Ia supernovae, we assess the viability of these models. Our analysis shows that CPL remains a strong and competitive parametrization, with the proposed alternatives only marginally favored or disfavored. We conclude that current observational data lack the statistical precision to decisively distinguish between CPL and similarly constructed parametrizations across the redshift range probed by late-time observables.