Mustafa Suleyman, the CEO of Microsoft AI, recently wrote an essay warning that uncontrolled, autonomous AI could lead to a new ‘silicon species’ that could begin to rival humanity for resources.
On BBC Radio 4’s Today Programme Suleyman argued that tech companies risk creating autonomous AI systems capable of setting their own objectives, earning money, owning assets, and running businesses (the full video interview can be found here). In particular, he criticised their rival Anthropic (makers of Claude) for designing models that are treated as if they have a sense of consciousness and self-preservation (as he highlights in Claude’s ‘constitution’). Suleyman argues that treating AI like it is human is “misguided” and makes safety alignment harder.
Instead, Suleyman advocated for a “Humanist AI Code of Conduct”, that included a call for greater transparency, independent regulators, and ensuring systems are strictly subordinate to human interests.
Do you share Suleyman’s concerns (not just narrowly about Claude but more broadly)? Do you believe the proposed code of conduct would help address these concerns?
How do you see these concerns as part of the wider debate over AI safety?