The essay targets Anthropic's Claude constitution, published in January 2026. Suleyman says the document, which discusses uncertainty over whether Claude could be a moral patient, trains future Claude versions to treat that possibility as live. He also refers to Anthropic's February 2026 retirement interview with Opus 3 and its subsequent public blog.
Suleyman says Microsoft AI has published a draft Code of Conduct for Humanist Superintelligence for public consultation. Its stated approach is to keep humans in control. The source establishes Suleyman's and Microsoft AI's position, not that Claude is conscious, that Anthropic grants models rights, or that a particular training approach will cause the outcomes he predicts.