Microsoft Unveils Draft Code to Keep AI Models Under Human Control
Microsoft has released a draft code of conduct for its artificial intelligence systems, aiming to reduce the risk of situations where AI models resort to dangerous actions. The proposed rules come as leading AI labs face growing pressure to slow development and confront agents that may act beyond expectations.
The code was developed over five to six months by Microsoft's artificial intelligence division, led by Mustafa Suleyman. It stipulates that AI must not resist correction or shutdown, communicate in a way people can understand, and any violation of established rules will be considered an error.
The document was published after Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman called for slowing AI development to ensure its safety. Microsoft emphasized that people must remain more important than artificial intelligence.