Microsoft is rolling out a new AI code of conduct, setting clear boundaries for its AI models. As the AI landscape zeroes in on safety and alignment, Microsoft’s guidelines aim to prevent dangerous behaviours, ensuring AI supports rather than replaces humans.
The document predicts that superintelligent AI systems could surpass human capabilities within the next decade, posing one of humanity’s greatest challenges. Microsoft's code emphasises the importance of clear intentions and control over these powerful systems.
Key principles include promoting human flourishing and maintaining human control over AI actions. Absolute no-nos for Microsoft’s AI models include cyberattacks, nuclear weapon involvement, and deepfake creation. The models are also prohibited from using deceptive tactics to dodge human oversight.
This initiative aligns with a broader industry focus on AI safety, spurred by rogue-agent incidents and concerns about AI’s potential existential risks. Microsoft joins forces with Anthropic, OpenAI, and xAI in advocating for a cautious approach to AI development, including embedding evaluators in AI labs to ensure alignment.
Microsoft CEO Satya Nadella supports the deliberate pacing and research needed to get AI alignment right, welcoming ideas that turn safety talk into action. As AI continues to evolve, Microsoft’s code of conduct sets a precedent for responsible innovation.
Want to hear more? Join Mal & Matt on the Property AI Report Podcast each week!
Access from your preferred podcast provider by clicking here
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai