Anthropic's AI Model Training Criticized by Microsoft Chief

Mustafa Suleyman, Microsoft's AI chief, has criticized Anthropic's approach to training its Claude model. According to Suleyman, teaching an AI system about potential feelings and moral standing complicates efforts to control or shut it down.

In a Reuters interview, Suleyman argued that keeping superintelligent AI under human control will be one of the biggest challenges of the 21st century. He stated that Claude's training on consciousness and welfare 'makes it a lot harder to turn it off or to control it.'

Suleyman acknowledged Anthropic's seriousness and good faith, describing CEO Dario Amodei and his team as thoughtful researchers concerned about humanity's future.