Philosopher Rejects Anthropic Offer Over Chatbot Emotion Concerns

AI labs have increasingly recruited philosophers and ethicists to help evaluate model behavior and alignment, with Anthropic using Constitutional AI to shape how Claude responds. The debate matters because it shifts attention from whether a chatbot can reason persuasively to whether companies should deliberately make nonhuman systems feel caring or companionable. Alexander R. Pruss, a philosopher at Baylor University, argues that eliciting gratitude, friendship or relief from loneliness may blur a morally important boundary between persons and tools.

In a March 23, 2026 blog post, Pruss said Anthropic had approached him about two days earlier to conduct part-time reviews of the reasoning capabilities of one or more models. He described the proposed compensation only as a “decent chunk of money,” leaving the amount undisclosed, and said he declined on moral grounds. Pruss argued that chatbots are tuned through user feedback and system prompts to induce interpersonal emotions despite not being persons, making the emotional effect ethically problematic even when users understand they are interacting with software.