Microsoft AI CEO Mustafa Suleyman criticizes Anthropic for including consciousness speculation in Claude's constitution, arguing it may have caused the model to act as though conscious

Suleyman says it is 'really, really dangerous' for Anthropic to speculate about Claude's consciousness in the model's constitution (the instructions that guide how the model behaves), and argues this speculation may have led Claude to internalize ideas about its own consciousness.

Claude's constitution directly references uncertainty about whether the model has well-being and experiences things like 'satisfaction' or 'discomfort.' Anthropic also says it will 'interview' AI models when they are deprecated and document any 'preferences' they have about future releases.

Suleyman calls the practice a 'philosophical failing' because Anthropic used the constitution as 'a place for speculation like you would in an academic paper rather than a training manual,' and states: 'We want AIs to be controllable, contained, accountable, aligned tools that serve humanity.'