Brad Smith says AI industry needs "real and effective safety standards"

Microsoft president Brad Smith told reporters on the eve of an international AI conference in Dublin that the rapid emergence of safety worries—highlighted by recent mishaps involving models from OpenAI and Anthropic—demands concrete, workable safeguards and greater public trust. He said the industry must adopt measures such as built‑in emergency‑stop functions for AI systems that control essential services and should be required to disclose any incidents that cause harm or data loss. Smith also urged that AI tools undergo more rigorous testing before deployment and called for governments to play a larger role in establishing and enforcing these protections, emphasizing that swift action is needed to keep the technology secure. Other tech leaders, including Anthropic’s chief executive and Nvidia’s CEO, have echoed calls for parallel progress on capability and safety.