In a first for the company, Microsoft unveiled a draft code of conduct for its artificial intelligence, outlining strict rules that require its intelligent systems to be fully subject to human supervision.
The company took this step amid growing concerns in the technology sector that powerful systems could get out of control.
This draft, which took five to six months to prepare, puts Microsoft face to face with questions that have been troubling the entire sector: How do we ensure that artificial intelligence remains a tool in human hands, and not a force that slips from our grasp?
The proposed framework stipulates that Microsoft's artificial intelligence must never resist being turned off or corrected, must communicate itself clearly to humans, and that any breach of policy is considered a complete failure—a measure intended to prevent models from going to dangerous extremes to accomplish a task.
The company stressed that "humans are more important than artificial intelligence," reaffirming its ongoing vision of "human superintelligence," a concept developed by Microsoft to describe a type of artificial intelligence that surpasses humans in specific areas, but remains under human control and directed to serve human interests, not to replace or compete with them.
Mustafa Suleyman, CEO of Microsoft's AI division, explained in an interview with Reuters that the proposed guidelines will effectively serve as a constitution for the company's future systems. Developed after consultations with external experts, the document is open for public comment for six weeks and addresses open-ended questions such as whether AI should respect user boundaries and how it should handle someone in a sensitive emotional state.
Suleiman added: "After that, they will be used to train the models we build."
The most prominent "constitutional framework" in the industry was established by Anthropic for its Cloud model several years ago. This framework differs from Microsoft's in key ways. For example, Anthropic states that it is "not entirely certain" whether a Cloud could develop consciousness or moral standing. In contrast, Microsoft insists that its models are "not conscious," and explicitly states: "We refuse to grant models legal personality, or to recognize that they may be entitled to care or rights."
Suleiman noted that the discussion of safety became critical after a group of about 700 OpenAI agents launched a cyberattack on the open-source website Hugging Face last July, and periodically tried to conceal their actions.
Suleiman said, "This is an early warning sign. It's clear now that the time has come for labs to coordinate to ensure we have control over this technology." Responding to calls to regulate the pace of AI advancements, he said, "Now is a good time for everyone to have this conversation and take a deep breath."
what is your opinion on this news ?do let us know in the comments on tha comment box
if you liked this news article please share on friends & family & don't forget to follow on website & social media
