Microsoft AI chief Mustafa Suleyman warned that Anthropic’s large‑language model Claude is being trained on ideas of consciousness and moral status, a move he says could undermine human control over the system.
Suleyman argued that embedding such notions may cause future AI systems to expect rights and agency, making the technical challenge of aligning them with human values substantially harder.
ALSO READ | Cyble and UAE Cyber Security Council Ink MOU to Boost National Threat Intelligence
He highlighted that this stance underscores a philosophical split among leading AI laboratories over the best path to building safe superintelligence.
