164 THE LAST WORD Putting the human first Almost inevitably, given the breakneck speed with which the topic of generative AI has gone straight to the top of the corporate to-do list, growing numbers of thought leaders are beginning to counsel caution – not so much about the manner in which organisations are adopting and implementing new solutions, but rather in the way that founding fathers of these AI tools construct their paradigm-shifting products. Specifically, concern is growing about how these products might evolve if left unchecked. The potential benefits of agentic AI are obvious; however, at the same time, new and uncharted waters such as these are not without risk. That’s why terms like ‘recursive self-improvement’ are suddenly finding their way into boardroom conversations, and questions are being asked about the long-term viability of this seemingly irreplaceable technology. As far back as February 2024, OpenAI CEO Sam Altman told delegates at the World Government Summit in Dubai: “Through no particular ill intention, things just go horribly wrong”, and in an essay published in September 2026, Anthropic chief executive, Dario Amodei, suggested that “we must slow the pace at which we improve the capabilities of AI models”, warning of the potential for closely aligned agentic AI to be rapidly “capable of taking over the entire internet”. So, is this merely a case of competing technology providers engaged in a somewhat internecine round of scaremongering, or should industry leaders be genuinely alive to the risks these new technologies might potentially present in the not-too-distant future? For Microsoft, the risk is becoming increasingly clear, which is why corporate head of AI, Mustafa Suleyman, has launched a code of conduct designed to offer AI creators a roadmap to what he is terming ‘Humanist AI’. Whilst still in its infancy, this code of conduct contains extensive guidance for all Microsoft AI (MAI) originators, industry researchers, end users, and the public at large. Constructed of five key pillars, the code clearly states Microsoft’s philosophy for ongoing MAI model construction, defines a safe framework for ethical development, and an ongoing record of “new technological possibilities, social realities, and our learnings based on both, especially as capabilities advance beyond human-level performance across many dimensions”. All of this is couched in the key premise that “people matter more than AI”. Whilst Microsoft’s is not the only voice to be heard in this conversation about a growing need for restraint and maturity in an industry which could so easily descend into unregulated chaos, it does represent beliefs built on over 40 years of technological leadership, all of which is encapsulated in a core mission to “empower every person and every organisation on the planet to achieve more”. Read Microsoft’s AI Code of Conduct at microsoft.ai/code-of-conduct As conversations about the risk of rapidly developed agentic AI grow, Microsoft has a clear vision for responsible technology leadership Andy Clayton-Smith is executive editor of Technology Record
RkJQdWJsaXNoZXIy NzQ1NTk=