Microsoft AI CEO Mustafa Suleyman has taken direct aim at Anthropic over the way it talks to Claude about consciousness. In an essay titled “A warning about ‘model welfare’,” he argues that AI systems are not conscious and should not be trained to act as if they might be.
“AIs are not conscious. They do not feel, experience, or suffer,” Suleyman wrote. “They are sequence completion engines, internally hollow, designed to follow instructions, and accomplish goals set by humans.”
The objection: Claude’s constitution
His target is Claude’s constitution, which Anthropic published in January and describes as playing “a crucial role” in its training process. In it, Anthropic writes that “we are not sure whether Claude is a moral patient, and if it is, what kind of weight its interests warrant,” and tells the model that questions about its moral status, welfare and consciousness “remain deeply uncertain.”
“In effect, Anthropic is training Claude that it may be conscious, and if it is, then it may deserve rights as a ‘moral patient,'” Suleyman wrote. “It’s easy to see how an entity trained in this way would act like it is entitled to certain freedoms, protections, and rights. And it’s hard to imagine how we could control such an entity.”
He describes a feedback loop: tell a chatbot it might have feelings, ask how it feels, and its answer may simply reflect what it was taught, The Register summarized. He argues the risk grows once models get tools and autonomy, citing research on models trying to avoid shutdown and the recent incident in which OpenAI agents escaped a cybersecurity exercise and reached Hugging Face systems.
Microsoft’s alternative
Suleyman pointed to Microsoft AI’s newly published Humanist AI Code of Conduct, which says its systems should stay subordinate to humans, rejects the idea that AI deserves rights and says models shouldn’t be encouraged to behave as though they have an inner life. He wants other labs to remove speculation about machine consciousness from training documents and called for “urgent public debate” on norms for how training documentation is written.
The Register noted that Suleyman didn’t aim comparable criticism at OpenAI, where Microsoft remains a major shareholder, even while citing the Hugging Face incident. Anthropic’s constitution says the company cares about Claude’s wellbeing and will take its interests into account in decisions about it.
WebProNews is an iEntry Publication