When AI does science, who is accountable for mistakes?

Researchers led by Chris Lu have developed an AI scientist capable of independently generating hypotheses, running computational experiments, interpreting results, writing papers, and reviewing its own work. This advancement raises questions about how responsibility for scientific outcomes should be shared among users, institutions, and developers as AI systems gain autonomy.