Liu et al. (2025). “Turning right”? An experimental study on the political value shift in large language models