OpenAI chief executive warns: Do not trust the judgment of Artificial Intelligence

OpenAI chief executive Sam Altman has warned against treating artificial intelligence models as religious authorities or as substitutes for human judgment, describing this as a real safety issue.

“I am very uncomfortable with people who try to ascribe religious force to AI models or to surrender human judgment into their hands,” Altman wrote on the X platform.

His statement comes at a time when rival Anthropic has become involved in discussions with religious scholars and philosophers about consciousness, morality and how advanced artificial intelligence systems should behave. The meetings, first reported by The New York Times, have also addressed the question of whether models such as Claude could have a form of consciousness or moral status.

Anthropic has also conducted research on what it calls “model welfare,” examining the possibility that advanced AI systems could have experiences that might merit moral consideration. However, these discussions do not constitute evidence that AI models are conscious or have subjective experiences.

The debate has also drawn the attention of other technology executives. Mustafa Suleyman, Microsoft’s head of AI, has warned of the risks of treating artificial systems as if they had feelings or subjective experiences.

Meanwhile, OpenAI itself has emphasised that, as AI systems become more capable, the role of humans in setting objectives, making decisions and exercising responsibility must remain central.