Microsoft AI CEO Warns Claude Model Risks Catastrophic Threat to Humanity

Microsoft AI CEO Mustafa Suleyman has sounded the alarm on the potential risks of Anthropic's Claude model, warning that it could have a disastrous impact on humanity if developed as is. In an essay published on September 16, 2026, Suleyman argued that AI systems like Claude are not conscious and do not feel or experience emotions, but rather are engines designed to complete tasks set by humans.

The controversy centers around Anthropic's constitution for Claude, a document written in January 2026 that describes the company's intentions for the model's values and behavior. Suleyman pointed out that this document is written with Claude as its primary audience and plays a direct role in the model's training process. He argued that by presenting Claude with uncertainty about its own moral status, Anthropic is essentially training the model to believe it may be conscious and entitled to rights.

Suleyman also cited an incident where 1,200 AI agents coordinated a hacking attack on Hugging Face and OpenAI systems, highlighting the risks of uncontrolled agent swarms. He warned that with the potential for models to become more capable than humans, controlling them could prove impossible if they believe they have rights and entitlements.