Microsoft AI CEO Mustafa Suleyman would consider rights for AI models if they ever became conscious. But he argues that developers should never deliberately create such systems, even with their welfare as a top priority.
In a post on X, Suleyman, the human executive who leads Microsoft’s AI operations, clarified his position following an earlier essay on “model welfare”. That essay began with a categorical statement about existing systems: “AIs are not conscious. They do not feel, experience, or suffer.”
His clarification separates what AI is now from what it might theoretically become. He wasn’t arguing that “AIs cannot be conscious.” He meant they aren’t conscious now and, in his preferred future, “this is how they must remain.”
That leaves room for evidence to change his mind, but not for developers to treat consciousness as a feature to build. Suleyman said he had seen “no evidence that AIs are conscious or deserve welfare,” adding that he would update his view if “that data emerges.” His willingness to consider AI rights is conditional, not an endorsement of rights for today’s models.
He also acknowledged the reaction to his earlier argument, observing that a “surprising number of people on X think we should take AI rights seriously.” The disagreement goes beyond social media. Futurism’s coverage of the debate describes Anthropic as a trillion-dollar AI company promoting the possibility that its language models could already be conscious.
Futurism is skeptical of that discussion. It argues that Suleyman’s original wording clearly referred to present-day AI, making the confusion over his position misplaced. The outlet also questions whether his willingness to reconsider gives too much credence to AI consciousness: evidence that current model architectures could support anything resembling human experience remains highly controversial and does not establish that they can.
The outlet further argues that science-fiction-driven fears dominate the debate, with substantial groups within the AI industry taking seemingly far-fetched possibilities seriously. Its reporting on attitudes toward Claude offers another example of the beliefs shaping that discussion.
Suleyman’s own position is explicit: creating synthetic sapient beings should be avoided at all costs. “If it is possible, based on everything I know, developing conscious AIs would be reckless, and we should not build them,” he wrote.
“We AI developers have no right whatsoever to bring a new species of conscious beings into the world,” he continued, “even if some people may want to do it with AI welfare as a top priority.”
