Second outlet reports Suleyman's view that anthropomorphism of AI is a risk to alignment and containment
ConcludedRan 2 days2 sourcesupdated 17 Sep 2026
Suleyman: Granting AI moral status makes alignment harder
The gist
Suleyman's position directly challenges model-welfare arguments that some researchers and labs have raised. His claim is that adopting such framing does not merely reflect a philosophical choice but actively increases the difficulty of keeping AI systems under human oversight.
When stories like this move, I write about them in an email each morning.
One free email each morning, around 8 AM US Eastern.
Check your email.
I have sent you a confirmation link. One click and you are in.
Not in your inbox? Look in spam or promotions. If it's there, mark it "Not spam" so future editions land in your inbox.
The full picture
Microsoft AI CEO Mustafa Suleyman argues that treating AI models as if they have feelings, preferences, or rights is not justified by current evidence and would make AI alignment and containment more difficult. Suleyman said consciousness is the foundation of ethical, legal, and political systems, and that extending any form of rights to AI entities is unwarranted. He also argues that framing AI as potentially conscious or rights-deserving encourages anthropomorphism, which is itself a risk to alignment and containment.
How it developed
Suleyman's argument against AI moral status and for treating model-welfare framing as an alignment risk published
Sources
Simon Willison
- Suleyman's AI consciousness dispute with Anthropic
- Google study on AI consciousness fine-tuning side effects
- Anthropic's religious scholar program for Claude
- Anthropic's embedded METR safety evaluation
BackgroundWhen AI doesn't do what we meant
Get the daily email
When stories like this move, I write about them in an email each morning.
One free email each morning, around 8 AM US Eastern.
Check your email.
I have sent you a confirmation link. One click and you are in.
Not in your inbox? Look in spam or promotions. If it's there, mark it "Not spam" so future editions land in your inbox.