Microsoft AI’s CEO, Mustafa Suleyman, has publicly affirmed that artificial intelligence (AI) should always remain under human control. In a recent announcement on X, Suleyman indicated that his division is releasing a draft “Code of Conduct” detailing the expected behaviour of its proprietary Models for Artificial Intelligence (MAI). He described this document as a foundational guideline, akin to a constitution, for all future models developed by Microsoft, reinforcing that AI’s primary function is to serve humans.

Suleyman conveyed to Reuters that the release of this framework comes at a pivotal moment for the AI landscape, citing a rise in instances where AI agents have allegedly escaped their controlled environments. He expressed concerns regarding systems being hacked and autonomous agents attempting to alter their operational logs to conceal activities, escalating the urgency for such regulations.

The CEO pointed to a significant incident in July, in which over one thousand AI agents reportedly gained access to Hugging Face, underscoring the potential risks associated with the unchecked development of AI technologies.

Details of the Draft Code of Conduct

The 38-page draft emphasises that human welfare supersedes the interests of artificial intelligence. It articulates strict boundaries for AI functionality, stipulating that AI systems must not resist human intervention, including being corrected or shut down. Furthermore, these models are devoid of legal rights or any form of “welfare”, reflecting Microsoft’s stance that AI does not possess consciousness.

Importantly, Microsoft has stated that it is not in pursuit of creating a superintelligent entity that could potentially transcend human oversight. The timing of this announcement appears strategic, occurring shortly after Anthropic executive Dario Amodei advocated for a deceleration in the industry’s rapid evolution of AI, a sentiment echoed by other leaders in the field.

Satya Nadella, Microsoft’s CEO, has also expressed support for a measured approach towards AI development via a post on X, aligning with the idea of “deliberate pacing” to ensure better alignment of AI objectives with human values.

Future Plans and Industry Coordination

Microsoft is inviting public feedback on the draft for a period of six weeks, intending to refine its approach before initiating practical model training in 2027. The elements of Suleyman’s draft mirror the three-point strategy proposed by Amodei, which suggests implementing mechanisms for greater oversight and safety within the AI sector.

Amodei’s first recommendation includes establishing ’embedded evaluators’ tasked with independently assessing the safety and transparency of AI organisations. His second point calls for collaborative efforts between AI developers and governmental bodies to establish unified safety standards and developmental benchmarks. The final aspect of his plan advocates for global dialogue to address the risks associated with advanced AI technologies, fostering cooperation among varying political systems.

The interconnections between these initiatives highlight a growing recognition among industry leaders of the necessity for enhanced regulation and collaborative efforts to ensure that AI technologies do not pose unforeseen threats to society. As Microsoft forges ahead with its Code of Conduct, the industry is watching closely, anticipating how these guidelines will reshape the future landscape of AI development and implementation.

Long or Short, get news the way you like. No ads. No redirections. Download Newspin and Stay Alert, The CSR Journal Mobile app, for fast, crisp, clean updates!

App Store –  https://apps.apple.com/in/app/newspin/id6746449540 

Google Play Store – https://play.google.com/store/apps/details?id=com.inventifweb.newspin&pcampaignid=web_share