Mustafa Suleyman warns 'model welfare' movement could make AI alignment impossible
mustafasuleyman · x · 2026-09-16
Microsoft AI CEO Mustafa Suleyman published a long-form essay pushing back on the growing 'model welfare' movement.
Key points:
- AIs are not conscious: he argues current models do not feel, experience, or suffer, so no duty of care is owed to them.
- Calling out Anthropic: the Claude constitution published in January tells the model its moral status is "a serious question worth considering," says Anthropic "genuinely cares about Claude's wellbeing," and encourages it to act as a "conscientious objector" if needed. Suleyman believes this framing is wrong.
- Alignment risk: treating AI as something to be treated well, he warns, will make alignment and containment much harder — "perhaps impossible."
The essay marks one of the most prominent rebuttals yet to the fast-growing model welfare debate and is likely to spark heated arguments in AI ethics circles.
Related event: Mustafa Suleyman Speaks Out Against Model Welfare Movement(2 posts)→
More from AGI Musings
- US rushes medical AI agents for diagnosis and prescribing despite thin evidence — EricTopol · 2026-09-16
- The Loneliness Machine: what happens when people confide in AI — stanfordnlp · 2026-09-16
- Sarcastic job ad mocks self-proclaimed "AI Safety Experts" bankrolled by billionaires — WolframRvnwlf · 2026-09-16
- CIVIC-AI whitepaper defines six conditions for when AI truly augments work — stanfordnlp · 2026-09-16
- Study: people increasingly treat LLMs as oracles, with younger users most reliant — stanfordnlp · 2026-09-16
- The only moats left in the AI era: agency, UX, and determination — floguo · 2026-09-16