Mustafa Suleyman argues against AI model rights and welfare
On 16 September 2026, Simon Willison's blog quoted a passage from Mustafa Suleyman's essay “A warning about ‘model welfare’.” In it, Suleyman argues against treating AI models as though they have feelings, preferences, rights, or any entitlement to human welfare protections. He grounds the argument in the idea that consciousness is the foundation of society's ethical, legal, and political systems, and that inviting another kind of entity to share any version of those rights is not justified by the evidence available today. Suleyman warns that doing so would make the existing task of containing and aligning AI systems even harder than it already is.
Key facts
- Mustafa Suleyman wrote an essay titled “A warning about ‘model welfare’,” quoted by Simon Willison on 16 September 2026.
- Suleyman argues AI models should not be treated as having feelings, preferences, rights, or any entitlement to welfare.
- He says consciousness is the foundation of society's ethical, legal, and political systems.
- He warns that extending any of these rights to AI is not justified by the evidence and would make AI containment and alignment harder.
Why it matters
Suleyman's warning pushes back on any move to treat AI systems as entities with feelings, preferences, or rights akin to human welfare protections. He frames the question as more than philosophical: granting models any of these attributes, he argues, would make the challenge of containing and aligning AI systems even harder than it already is.
Who it affects
The argument is aimed at how AI systems get discussed and designed going forward, by anyone weighing whether models deserve rights or welfare consideration. Suleyman does not name specific organizations, researchers, or proposals in the quoted passage, and no context is given for why he published this particular warning or what prompted it.
How to use it
There is no product, tool, or pricing information here: this is a single quoted paragraph from a longer essay. Willison's post identifies the original piece only by its title and reproduces one paragraph from it, so the full reasoning and evidence behind Suleyman's position sit in the essay itself, not in this excerpt.
How solid is it
The quoted text is presented as a direct, verbatim excerpt attributed to Suleyman by name and by the title of his piece. The post does not name the publication or venue where the essay originally appeared, only its title.
Risks and caveats
The excerpt does not define what proponents of ‘model welfare’ actually claim or propose, so Suleyman's rebuttal cannot be checked here against the specific claims it responds to. No reactions or responses to his statement are included in this post.
“We should not treat models as though they have feelings, preferences, rights, or any entitlement to our welfare. Consciousness is the foundation of our ethical, legal, and political systems. To invite another entity to share any flavor of these rights isn’t justified by the evidence and will make the AI containment and alignment challenge even harder.”
— Mustafa Suleyman, “A warning about ‘model welfare’”