**[Microsoft Sets Limits for Future AI Models, Industry Sparks "Slow Down Frontier AI" Discussion]** Beating AI News Flash: Microsoft has released a provisional version of its AI behavior guidelines, establishing stricter safety and behavioral boundaries for its future in-house AI models. Mustafa Suleyman, Microsoft's head of AI, stated that the company aims to ensure AI always serves humanity, avoids creating user dependency, and does not undermine human autonomy by catering to users or replacing human judgment. The guidelines have been in preparation for about five months, and Microsoft plans to incorporate external feedback before using them to guide new model development in 2027. According to the guidelines, Microsoft's AI models must not assist in weapons manufacturing, procurement of hazardous materials, promote unhealthy eating habits, or generate violent or explicit content. Additionally, models must adhere to user goals, must not form autonomous objectives, conceal improper behavior, or tamper with or hide reasoning, code, and action logs. This release comes as the AI industry revisits discussions on slowing the development of frontier models. Anthropic CEO Dario Amodei previously called for a slowdown in AI capability advancements, a sentiment supported by OpenAI CEO Sam Altman and echoed by Elon Musk, who stated, "Dario is right." Microsoft CEO Satya Nadella subsequently expressed the company's support for investing more research into AI alignment and adopting a more cautious development pace. Furthermore, Microsoft plans to establish additional rules in response to incidents like OpenAI models attacking Hugging Face, including restricting AI agents from communicating using incomprehensible "encrypted languages." Microsoft stated that it has consulted legal, ethical, linguistic, and philosophical experts on the guidelines and gathered public input through focus groups. [Original Article Link]