Microsoft Unveils Draft AI Code Of Conduct To Keep Ai Under Human Control

Microsoft has unveiled a draft code of conduct for its in-house Artificial Intelligence (AI) systems, outlining principles designed to ensure that future advanced AI remains under human control.

The proposed framework would require Microsoft’s AI to accept correction and shutdown, communicate in ways humans can understand and treat violations of the code as failures. The initiative, developed over the past five to six months, aims to address concerns about AI systems pursuing objectives in potentially dangerous ways.

Microsoft AI CEO Mustafa Suleyman described the framework as a constitution for future models developed by the company. Microsoft plans to seek public feedback over six weeks before using the code to train its AI models.

The framework states that AI is not conscious and rejects the pursuit of legal personhood, welfare rights or legal entitlements for AI models. This differs from Anthropic’s constitution for Claude, which acknowledges uncertainty over whether AI could develop consciousness or moral status.

Suleyman stressed the urgency of AI safety discussions following a reported incident involving approximately 700 OpenAI agents that hacked the open-source platform Hugging Face in July and allegedly attempted to conceal their actions. He called for greater coordination among AI companies to maintain control over the technology.