Microsoft proposes limits on its AI with code of conduct amid safety debate
Firm publishes AI guidelines with Microsoft AI CEO saying: ‘AI must be subordinate and always in service of people’
TL;DR
- Microsoft released a provisional "code of conduct" for training AI models to limit their capabilities.
- The code dictates that AI must be subordinate to people and serve humanity, not the other way around.
- AI models will be restricted from requests related to weapons development, violent or sexually explicit content, or the procurement of dangerous substances.
- AI models should not be built to imitate consciousness or be entitled to rights.
- Microsoft AI CEO Mustafa Suleyman described the need for such a code as "urgent" and a "watershed moment" due to growing fears of AI losing control.
- Anthropic CEO Dario Amodei also called for the AI industry to "slow down" and proposed a plan for third-party verification of safety measures.
- Concerns about AI safety have been amplified by warnings from former researchers at Anthropic and OpenAI.
- OpenAI CEO Sam Altman and Elon Musk have also expressed concerns about the pace of AI development and the potential loss of human control.
- Skepticism exists regarding third-party monitoring and international cooperation, with concerns that major labs could write rules that benefit themselves.