Matthew Hutson. "AI models that lie, cheat and plot murder: how dangerous are LLMs really?." Nature (2025). https://doi.org/10.1038/d41586-025-03222-1