Microsoft's brad smith advocates for independent AI safety assessments and human control

At the United Nations General Assembly in New York, Microsoft President Brad Smith called for independent assessments of AI safety, emphasizing the need for systems to remain under human control. During an interview with Bloomberg on September 23, Smith argued that relying on just a few institutions to oversee AI safety would hinder progress. He highlighted the importance of companies like Microsoft, which provide the software layer and monitor AI behavior. Microsoft collaborates with OpenAI through a joint safety committee, where both parties contribute members to evaluate AI safety. Smith also supported the inclusion of a 'kill switch' for advanced systems, ensuring human oversight. He acknowledged that Microsoft needs to revise its climate goals, as its current target of removing more carbon than emitted by 2030 appears increasingly unattainable due to reliance on gas-powered data centers. Smith admitted that adjustments are necessary to meet these goals, stating that the approach must evolve from what was envisioned six years ago.