Microsoft’s Mustafa Suleyman Walks Back Automation Claims and Slams Anthropic Over Anthropomorphizing Claude

During a recent appearance on The Verge's Decoder podcast, Microsoft AI CEO Mustafa Suleyman sought to reframe the conversation around artificial intelligence in the workplace, walking back his earlier, more alarming prediction about automation while also opening a new front in the philosophical battle over how AI should be described to the public .

The task-versus-job distinction

Suleyman's most immediate goal was to clarify a statement he made to the Financial Times months earlier, which was widely interpreted as a prediction that most white-collar jobs would be automated within 12 to 18 months . On Decoder, he argued that this was a misreading of his intent . The operative word, he now insists, was always "tasks." He was not forecasting the wholesale elimination of roles like lawyers, accountants, or project managers, but rather the automation of the computer-based sub-tasks that consume their days .

He gave concrete examples of the digitized work to come: "sending an email, having a conversation with colleagues, putting together PowerPoints" . In his view, this wave of automation will make knowledge workers faster and more efficient, not render them redundant .

"That does not necessarily mean that the role goes away at all. It just means that the work can be done faster and more efficiently, which is today often work that is quite rote, is quite manual, is quite labor-intensive, and is time-consuming."

This places Suleyman firmly in the camp of pragmatic task-level automation, defusing more dramatic doomsday predictions about white-collar labor .

A shot across the bow at Anthropic

Suleyman used the same podcast appearance to level a clear criticism at rival AI lab Anthropic. According to the episode’s description, he took issue with the company "talking about Claude as though it is conscious" . The pointed remark is not about the raw capability or safety of Anthropic's models, but about the language and framing used to present them to the world .

The critique draws a sharp line between two philosophies: one that treats AI as an instrumental software tool, and another that, in Suleyman’s view, risks anthropomorphizing systems in a way that implies human-like awareness or personhood .

The twin fault lines exposed

Suleyman’s Decoder comments reveal two interconnected debates currently fragmenting the AI industry:

• Automation realism vs. job doomsaying: By clarifying his "tasks" prediction, Suleyman is distancing himself from a simplistic narrative that AI will soon decimate white-collar employment. His position acknowledges massive near-term productivity gains without buying into the most drastic labor-market forecasts .
• Instrumental framing vs. anthropomorphism: His swipe at Anthropic over Claude's portrayal is not a one-off comment. It reflects a growing tension over how AI labs choose to describe their creations. Framing a model as conscious or person-like carries enormous cultural, ethical, and regulatory weight. Suleyman is drawing a line in the sand, advocating for language that treats AI as advanced software rather than a nascent mind .

Together, these interventions paint a picture of a Microsoft AI chief who sees superintelligence on the horizon but is determined to keep the conversation grounded in practical tool-making rather than speculative philosophy .