Microsoft Warns Against Unchecked AI Progress with New Guiding Principles

Microsoft has joined a growing list of AI firms calling for caution in developing cutting-edge models. The company's researchers have released a set of guiding tenets, including the core principle that 'people matter more than AI.' The document, which was months in the making, outlines specific limits on Microsoft's development of advanced AI models.

The 15,000-word manifesto emphasizes that AI models should not be given rights or legal personhood. They should also not be engineered to escape human control or deceive users. If completing a task requires violating these principles, the system should not complete the task.

This is in response to recent security incidents that have raised concerns about the existential risks of advanced AI technology. Microsoft's Humanist AI 'rejects the race to produce an all-purpose superintelligence' and prioritizes building something fundamentally useful and safe.

The guidelines go further than those set by OpenAI and Anthropic, including prohibitions on anthropomorphizing AI and resisting human control over a model's process.